{"cells":[{"metadata":{"_uuid":"e8e112aa5b5db7303e39d3a9cf1702848962c6c2"},"cell_type":"markdown","source":"### Introduction\n\nUsing Olivier's wonderful script as a base line I decided to evolve models using GP. \nNote that I made it a logistic regression problem by scaling outputs to between zero and one before evolving the models.  (I don't like minimizing to zero if the score is less than zero!)\n"},{"metadata":{"_uuid":"8f2839f25d086af736a60e9eeb907d3b93b6e0e5","_cell_guid":"b1076dfc-b9ad-4769-8c92-a6c4dae69d19","trusted":true},"cell_type":"code","source":"import os\nprint(os.listdir(\"../input\"))","execution_count":null,"outputs":[]},{"metadata":{"_cell_guid":"79c7e3d0-c299-4dcb-8224-4455121ee9b0","_uuid":"d629ff2d2480ee46fbb7e2d37f6b5fab8052498a","trusted":true},"cell_type":"code","source":"import pandas as pd\nimport numpy as np\nimport matplotlib.pyplot as plt\nimport seaborn as sns\nfrom sklearn.metrics import mean_squared_error\nimport gc\nimport time\nfrom pandas.core.common import SettingWithCopyWarning\nimport warnings\nimport lightgbm as lgb\nfrom sklearn.model_selection import GroupKFold\n\n# I don't like SettingWithCopyWarnings ...\nwarnings.simplefilter('error', SettingWithCopyWarning)\ngc.enable()\n%matplotlib inline","execution_count":null,"outputs":[]},{"metadata":{"_uuid":"e8c6c0d5d8cf00737b8ae536b0d528b32fac9166"},"cell_type":"markdown","source":"### Get the extracted data"},{"metadata":{"trusted":true,"_uuid":"6fe04db087c3d25b8b4a5873ee6f104e427525f0"},"cell_type":"code","source":"train = pd.read_csv('../input/create-extracted-json-fields-dataset/extracted_fields_train.gz', \n                    dtype={'date': str, 'fullVisitorId': str, 'sessionId':str}, nrows=None)\ntest = pd.read_csv('../input/create-extracted-json-fields-dataset/extracted_fields_test.gz', \n                   dtype={'date': str, 'fullVisitorId': str, 'sessionId':str}, nrows=None)\ntrain.shape, test.shape","execution_count":null,"outputs":[]},{"metadata":{"_uuid":"b64b336b70ccb5478d33f9e042eeeabee549c919"},"cell_type":"markdown","source":"### Define folding strategy"},{"metadata":{"trusted":true,"_uuid":"cd54722ebf9ae8b68eddb2116c0559625e7d6a48"},"cell_type":"code","source":"def get_folds(df=None, n_splits=5):\n    \"\"\"Returns dataframe indices corresponding to Visitors Group KFold\"\"\"\n    # Get sorted unique visitors\n    unique_vis = np.array(sorted(df['fullVisitorId'].unique()))\n\n    # Get folds\n    folds = GroupKFold(n_splits=n_splits)\n    fold_ids = []\n    ids = np.arange(df.shape[0])\n    for trn_vis, val_vis in folds.split(X=unique_vis, y=unique_vis, groups=unique_vis):\n        fold_ids.append(\n            [\n                ids[df['fullVisitorId'].isin(unique_vis[trn_vis])],\n                ids[df['fullVisitorId'].isin(unique_vis[val_vis])]\n            ]\n        )\n\n    return fold_ids","execution_count":null,"outputs":[]},{"metadata":{"_uuid":"18b2fb7729077132ef249956a1bbb072baee17db"},"cell_type":"markdown","source":"### Get session target"},{"metadata":{"trusted":true,"_uuid":"c539c76d9a5305066ea084ca32016132da41b35b"},"cell_type":"code","source":"y_reg = train['totals.transactionRevenue'].fillna(0)\ndel train['totals.transactionRevenue']\n\nif 'totals.transactionRevenue' in test.columns:\n    del test['totals.transactionRevenue']","execution_count":null,"outputs":[]},{"metadata":{"_uuid":"1d9af86b1cf94e192d704bfd419eed4490e1debe"},"cell_type":"markdown","source":"### Add date features\n\nOnly add the one I think can ganeralize"},{"metadata":{"trusted":true,"_uuid":"fbf93fda18940aa29105fa2d66f3514311d4fc68"},"cell_type":"code","source":"train.columns","execution_count":null,"outputs":[]},{"metadata":{"trusted":true,"_uuid":"b04391e12fdca0199ff22ab7febd9b0b53d6cbb3"},"cell_type":"code","source":"for df in [train, test]:\n    df['date'] = pd.to_datetime(df['visitStartTime'], unit='s')\n    df['sess_date_dow'] = df['date'].dt.dayofweek\n    df['sess_date_hours'] = df['date'].dt.hour\n    df['sess_date_dom'] = df['date'].dt.day","execution_count":null,"outputs":[]},{"metadata":{"_uuid":"59e7c9bcf14d3f430489d15d22757b91dd9d5ad1"},"cell_type":"markdown","source":"### Create features list"},{"metadata":{"trusted":true,"_uuid":"e09ff3124721ad70eab65b924093b5fd0c12f9c6"},"cell_type":"code","source":"excluded_features = [\n    'date', 'fullVisitorId', 'sessionId', 'totals.transactionRevenue', \n    'visitId', 'visitStartTime'\n]\n\ncategorical_features = [\n    _f for _f in train.columns\n    if (_f not in excluded_features) & (train[_f].dtype == 'object')\n]","execution_count":null,"outputs":[]},{"metadata":{"_uuid":"2d207a80600a44e758b6011f24c5808f9b164866"},"cell_type":"markdown","source":"### Factorize categoricals"},{"metadata":{"trusted":true,"_uuid":"4f00eb6b6166cd5be25ebad22aa9dc91a8a3ded2"},"cell_type":"code","source":"for f in categorical_features:\n    train[f], indexer = pd.factorize(train[f])\n    test[f] = indexer.get_indexer(test[f])","execution_count":null,"outputs":[]},{"metadata":{"_uuid":"d7517b473bd776f51925ab0243cf1e3b6592ce0e"},"cell_type":"markdown","source":"### Predict revenues at session level"},{"metadata":{"trusted":true,"_uuid":"e9c155197545dc43ec2f3e23e25955532f7ad25d"},"cell_type":"code","source":"folds = get_folds(df=train, n_splits=5)\n\ntrain_features = [_f for _f in train.columns if _f not in excluded_features]\nprint(train_features)\n\nimportances = pd.DataFrame()\noof_reg_preds = np.zeros(train.shape[0])\nsub_reg_preds = np.zeros(test.shape[0])\nfor fold_, (trn_, val_) in enumerate(folds):\n    trn_x, trn_y = train[train_features].iloc[trn_], y_reg.iloc[trn_]\n    val_x, val_y = train[train_features].iloc[val_], y_reg.iloc[val_]\n    \n    reg = lgb.LGBMRegressor(\n        num_leaves=31,\n        learning_rate=0.03,\n        n_estimators=1000,\n        subsample=.9,\n        colsample_bytree=.9,\n        random_state=1\n    )\n    reg.fit(\n        trn_x, np.log1p(trn_y),\n        eval_set=[(val_x, np.log1p(val_y))],\n        early_stopping_rounds=50,\n        verbose=100,\n        eval_metric='rmse'\n    )\n    imp_df = pd.DataFrame()\n    imp_df['feature'] = train_features\n    imp_df['gain'] = reg.booster_.feature_importance(importance_type='gain')\n    \n    imp_df['fold'] = fold_ + 1\n    importances = pd.concat([importances, imp_df], axis=0, sort=False)\n    \n    oof_reg_preds[val_] = reg.predict(val_x, num_iteration=reg.best_iteration_)\n    oof_reg_preds[oof_reg_preds < 0] = 0\n    _preds = reg.predict(test[train_features], num_iteration=reg.best_iteration_)\n    _preds[_preds < 0] = 0\n    sub_reg_preds += np.expm1(_preds) / len(folds)\n    \nmean_squared_error(np.log1p(y_reg), oof_reg_preds) ** .5","execution_count":null,"outputs":[]},{"metadata":{"_uuid":"2b23cdb7ee1e35c6deba9edd6ffd87a4cdc72920"},"cell_type":"markdown","source":"### Display feature importances"},{"metadata":{"trusted":true,"_uuid":"61d4aacc57c2c7c4d9ce0b1cab7221ca7c69138f"},"cell_type":"code","source":"import warnings\nwarnings.simplefilter('ignore', FutureWarning)\n\nimportances['gain_log'] = np.log1p(importances['gain'])\nmean_gain = importances[['gain', 'feature']].groupby('feature').mean()\nimportances['mean_gain'] = importances['feature'].map(mean_gain['gain'])\n\nplt.figure(figsize=(8, 12))\nsns.barplot(x='gain_log', y='feature', data=importances.sort_values('mean_gain', ascending=False))","execution_count":null,"outputs":[]},{"metadata":{"_uuid":"c64447b5858dc899b6b78d26a4e68a8582995051"},"cell_type":"markdown","source":"### Create user level predictions"},{"metadata":{"trusted":true,"_uuid":"1fc92504b40407cff766489db1966e729d0466df"},"cell_type":"code","source":"train['predictions'] = np.expm1(oof_reg_preds)\ntest['predictions'] = sub_reg_preds","execution_count":null,"outputs":[]},{"metadata":{"trusted":true,"_uuid":"515d3d14483844c93710fb6b6f9e39a6a2a1d384"},"cell_type":"code","source":"# Aggregate data at User level\ntrn_data = train[train_features + ['fullVisitorId']].groupby('fullVisitorId').mean()","execution_count":null,"outputs":[]},{"metadata":{"trusted":true,"_uuid":"31d76001e77e3ee2e10e67a5bc899a83c4cfeb8d"},"cell_type":"code","source":"%%time\n# Create a list of predictions for each Visitor\ntrn_pred_list = train[['fullVisitorId', 'predictions']].groupby('fullVisitorId')\\\n    .apply(lambda df: list(df.predictions))\\\n    .apply(lambda x: {'pred_'+str(i): pred for i, pred in enumerate(x)})","execution_count":null,"outputs":[]},{"metadata":{"trusted":true,"_uuid":"b8f1bac1ef32a58f70aa7bcca79acc7119ddcfb5"},"cell_type":"code","source":"# Create a DataFrame with VisitorId as index\n# trn_pred_list contains dict \n# so creating a dataframe from it will expand dict values into columns\ntrn_all_predictions = pd.DataFrame(list(trn_pred_list.values), index=trn_data.index)\ntrn_feats = trn_all_predictions.columns\ntrn_all_predictions['t_mean'] = np.log1p(trn_all_predictions[trn_feats].mean(axis=1))\ntrn_all_predictions['t_median'] = np.log1p(trn_all_predictions[trn_feats].median(axis=1))\ntrn_all_predictions['t_sum_log'] = np.log1p(trn_all_predictions[trn_feats]).sum(axis=1)\ntrn_all_predictions['t_sum_act'] = np.log1p(trn_all_predictions[trn_feats].fillna(0).sum(axis=1))\ntrn_all_predictions['t_nb_sess'] = trn_all_predictions[trn_feats].isnull().sum(axis=1)\nfull_data = pd.concat([trn_data, trn_all_predictions], axis=1)\ndel trn_data, trn_all_predictions\ngc.collect()\nfull_data.shape","execution_count":null,"outputs":[]},{"metadata":{"trusted":true,"_uuid":"12a86e1f61f819417bede61f3746366b5028222f"},"cell_type":"code","source":"%%time\nsub_pred_list = test[['fullVisitorId', 'predictions']].groupby('fullVisitorId')\\\n    .apply(lambda df: list(df.predictions))\\\n    .apply(lambda x: {'pred_'+str(i): pred for i, pred in enumerate(x)})","execution_count":null,"outputs":[]},{"metadata":{"trusted":true,"_uuid":"611f2a86ffff5947b6863818538577b58f099535"},"cell_type":"code","source":"sub_data = test[train_features + ['fullVisitorId']].groupby('fullVisitorId').mean()\nsub_all_predictions = pd.DataFrame(list(sub_pred_list.values), index=sub_data.index)\nfor f in trn_feats:\n    if f not in sub_all_predictions.columns:\n        sub_all_predictions[f] = np.nan\nsub_all_predictions['t_mean'] = np.log1p(sub_all_predictions[trn_feats].mean(axis=1))\nsub_all_predictions['t_median'] = np.log1p(sub_all_predictions[trn_feats].median(axis=1))\nsub_all_predictions['t_sum_log'] = np.log1p(sub_all_predictions[trn_feats]).sum(axis=1)\nsub_all_predictions['t_sum_act'] = np.log1p(sub_all_predictions[trn_feats].fillna(0).sum(axis=1))\nsub_all_predictions['t_nb_sess'] = sub_all_predictions[trn_feats].isnull().sum(axis=1)\nsub_full_data = pd.concat([sub_data, sub_all_predictions], axis=1)\ndel sub_data, sub_all_predictions\ngc.collect()\nsub_full_data.shape","execution_count":null,"outputs":[]},{"metadata":{"_uuid":"c901512b911c731113c0515e97ce80ddf7b7f118"},"cell_type":"markdown","source":"### Create target at Visitor level"},{"metadata":{"trusted":true,"_uuid":"bf3b05a18f83821d12d05500f8e0e52f0603d6c5"},"cell_type":"code","source":"train['target'] = y_reg\ntrn_user_target = train[['fullVisitorId', 'target']].groupby('fullVisitorId').sum()","execution_count":null,"outputs":[]},{"metadata":{"trusted":true,"_uuid":"325be2052c53d573832ab7cf2ab50397e267539b"},"cell_type":"code","source":"df = pd.concat([full_data,sub_full_data],sort=False)\ntrainsize = full_data.shape[0]\ndel full_data, sub_full_data\ngc.collect()\ndf = df.reset_index(drop=False)\nfor c in df.columns[1:]:\n    if((df[c].min()>=0)&(df[c].max()>=10)):\n        df[c] = np.log1p(df[c])\n    elif((df[c].min()<0)&((df[c].max()-df[c].min())>=10)):\n        df.loc[df[c]!=0,c] = np.sign(df.loc[df[c]!=0,c])*np.log(np.abs(df.loc[df[c]!=0,c]))\nfrom sklearn.preprocessing import StandardScaler\nfor c in df.columns[1:]:\n    ss = StandardScaler()\n    df.loc[~np.isfinite(df[c]),c] = np.nan\n    df.loc[~df[c].isnull(),c] = ss.fit_transform(df.loc[~df[c].isnull(),c].values.reshape(-1,1))\ndf.fillna(-99999,inplace=True)\ngp_trn_users = df[:trainsize].copy().set_index('fullVisitorId')\nmaxval = np.log1p(trn_user_target['target'].values).max()\ngp_trn_users['target'] = np.log1p(trn_user_target['target'].values)\ngp_trn_users.target /= maxval\ngp_sub_users = df[trainsize:].copy().set_index('fullVisitorId')\nnewcols =  [x.replace('.','_') for x in gp_trn_users.columns]\ngp_trn_users.columns = newcols\nnewcols =  [x.replace('.','_') for x in gp_sub_users.columns]\ngp_sub_users.columns = newcols\ndel df\ngc.collect()","execution_count":null,"outputs":[]},{"metadata":{"trusted":true,"_uuid":"ce401ec87770615d8b47b69b6ff498cb251bf1b5"},"cell_type":"code","source":"maxval ","execution_count":null,"outputs":[]},{"metadata":{"trusted":true,"_uuid":"cc00e6815558df9e7be10f2f86b877c7455ef2b7"},"cell_type":"code","source":"def Output(p):\n    return 1.0/(1+np.exp(-p))\n\ndef GP1(data):\n    v = pd.DataFrame()\n    v[\"i0\"] = 0.100000*np.tanh((((((np.tanh((((data[\"pred_4\"]) + (data[\"pred_6\"]))/2.0)) )) * ((12.07693195343017578)))) - ((10.0))))  \n    v[\"i1\"] = 0.100000*np.tanh(((np.where((np.tanh((((((((((10.0)) - (data[\"pred_5\"]))) + (2.0))) - (data[\"pred_0\"]))) * (3.0))) )<0, 3.0, ((-3.0) - ((10.0))) )) * ((((9.80441665649414062)) * 2.0))))  \n    v[\"i2\"] = 0.100000*np.tanh((((((np.tanh(((((-1.0) / 2.0)) + (data[\"pred_8\"]))) )) + (np.maximum((((np.tanh(((((-1.0) / 2.0)) + (data[\"pred_8\"]))) ))), ((-1.0)))))) * ((6.0))))  \n    v[\"i3\"] = 0.100000*np.tanh((((np.tanh(((-2.0) + (data[\"pred_3\"]))) )) * ((((-1.0*(((((3.0) + (((((-3.0) - ((9.0)))) * 2.0)))/2.0))))) * 2.0))))  \n    v[\"i4\"] = 0.100000*np.tanh((((np.tanh(((data[\"pred_3\"]) - ((2.44455146789550781)))) )) * ((10.55293273925781250))))  \n    v[\"i5\"] = 0.100000*np.tanh(((np.where((np.tanh((((data[\"pred_2\"]) + (-3.0))/2.0)) )>0, -3.0, (9.0) )) * (((((((-3.0) - ((9.0)))) * 2.0)) * 2.0))))  \n    v[\"i6\"] = 0.100000*np.tanh((((((((np.tanh(((((((2.48090791702270508)) + (data[\"pred_4\"]))/2.0)) + (-2.0))) )) * ((10.93176364898681641)))) * 2.0)) * 2.0))  \n    v[\"i7\"] = 0.100000*np.tanh((((((((np.tanh(((data[\"pred_1\"]) - (2.0))) )) * 2.0)) / 2.0)) * ((9.0))))  \n    v[\"i8\"] = 0.100000*np.tanh((((((np.where(((-3.0) - ((9.55589389801025391))) < -99998, -1.0, 1.0 )) > ((np.tanh(((data[\"pred_0\"]) - (-1.0))) )))*1.)) * (((-3.0) * 2.0))))  \n    v[\"i9\"] = 0.100000*np.tanh((((12.06137466430664062)) * ((np.tanh((((data[\"t_mean\"]) + (((-3.0) + (-2.0))))/2.0)) ))))  \n    v[\"i10\"] = 0.100000*np.tanh(((-3.0) * ((((10.0)) * ((np.tanh((((8.0)) - (data[\"t_mean\"]))) ))))))  \n    v[\"i11\"] = 0.100000*np.tanh((((((-1.0*(((((5.0)) * 2.0))))) / 2.0)) + (((np.where((np.tanh(((data[\"pred_2\"]) - (3.0))) )<0, (np.tanh(((data[\"pred_2\"]) - (3.0))) ), (6.70732259750366211) )) * 2.0))))  \n    v[\"i12\"] = 0.100000*np.tanh((((np.tanh(((((((-1.0*((3.0)))) * 2.0)) + (np.maximum(((-3.0)), ((data[\"t_mean\"])))))/2.0)) )) * ((13.89055156707763672))))  \n    v[\"i13\"] = 0.100000*np.tanh((((7.0)) * (np.where(np.minimum((((5.90891504287719727))), (((((np.tanh(np.minimum(((((data[\"t_mean\"]) - ((4.69426012039184570))))), (((((((-3.0) + (data[\"t_mean\"]))) + (np.maximum(((data[\"pred_5\"])), ((-3.0)))))/2.0))))) )) * ((np.tanh(np.minimum(((((data[\"t_mean\"]) - ((4.69426012039184570))))), (((((((-3.0) + (data[\"t_mean\"]))) + (np.maximum(((data[\"pred_5\"])), ((-3.0)))))/2.0))))) ))))))>0, (np.tanh(np.minimum(((((data[\"t_mean\"]) - ((4.69426012039184570))))), (((((((-3.0) + (data[\"t_mean\"]))) + (np.maximum(((data[\"pred_5\"])), ((-3.0)))))/2.0))))) ), (np.tanh(np.minimum(((((data[\"t_mean\"]) - ((4.69426012039184570))))), (((((((-3.0) + (data[\"t_mean\"]))) + (np.maximum(((data[\"pred_5\"])), ((-3.0)))))/2.0))))) ) ))))  \n    v[\"i14\"] = 0.100000*np.tanh((((11.24391651153564453)) * ((((np.tanh(((((((data[\"t_sum_log\"]) + ((((-2.0) + ((((-1.0) + (-2.0))/2.0)))/2.0)))/2.0)) + (((np.tanh((-2.0))) * 2.0)))/2.0)) )) * 2.0))))  \n    v[\"i15\"] = 0.100000*np.tanh(((np.where((7.0)>0, (np.tanh(((((((((data[\"t_sum_log\"]) / 2.0)) < (data[\"pred_0\"]))*1.)) + (((data[\"t_sum_log\"]) - ((6.0)))))/2.0)) ), (((((np.tanh(((((((((data[\"t_sum_log\"]) / 2.0)) < (data[\"pred_0\"]))*1.)) + (((data[\"t_sum_log\"]) - ((6.0)))))/2.0)) )) * ((((9.0)) * ((np.tanh(((((((((data[\"t_sum_log\"]) / 2.0)) < (data[\"pred_0\"]))*1.)) + (((data[\"t_sum_log\"]) - ((6.0)))))/2.0)) )))))) * 2.0) )) * ((6.0))))  \n    v[\"i16\"] = 0.100000*np.tanh((((((np.tanh(np.tanh((((1.0) - ((-1.0*((0.0)))))))) )) - ((9.39388847351074219)))) * ((((((((np.tanh((-1.0*((np.maximum(((-3.0)), ((((data[\"t_sum_log\"]) / 2.0)))))))) )) - (((-2.0) / 2.0)))) * 2.0)) * 2.0))))  \n    v[\"i17\"] = 0.100000*np.tanh(((((np.where((np.tanh((((-3.0) + (((data[\"t_sum_act\"]) / 2.0)))/2.0)) )>0, (np.tanh((((-3.0) + (((data[\"t_sum_act\"]) / 2.0)))/2.0)) ), (((((((np.tanh((((-3.0) + (((data[\"t_sum_act\"]) / 2.0)))/2.0)) )) + ((np.tanh((((-3.0) + (((data[\"t_sum_act\"]) / 2.0)))/2.0)) )))) * 2.0)) + ((1.0))) )) * 2.0)) * 2.0))  \n    v[\"i18\"] = 0.100000*np.tanh((((((((12.23419475555419922)) * ((np.tanh(((np.maximum(((((-3.0) + (data[\"pred_3\"])))), ((-3.0)))) + (((data[\"t_sum_log\"]) + (-3.0))))) )))) * 2.0)) - ((((((9.04281616210937500)) + ((9.0)))) * ((np.tanh(((np.maximum(((((-3.0) + (data[\"pred_3\"])))), ((-3.0)))) + (((data[\"t_sum_log\"]) + (-3.0))))) ))))))  \n    v[\"i19\"] = 0.100000*np.tanh((((((np.tanh(((data[\"t_sum_log\"]) - (((np.maximum((((7.0))), ((data[\"t_sum_log\"])))) / 2.0)))) )) * 2.0)) * 2.0))  \n    v[\"i20\"] = 0.100000*np.tanh((((np.tanh(np.minimum(((data[\"pred_5\"])), ((data[\"totals_pageviews\"])))) )) * ((13.97667598724365234))))  \n    v[\"i21\"] = 0.100000*np.tanh(((((np.tanh(np.maximum(((data[\"pred_4\"])), ((((data[\"t_sum_log\"]) + (-3.0)))))) )) + ((((((6.0)) * ((np.tanh(np.maximum(((data[\"pred_4\"])), ((((data[\"t_sum_log\"]) + (-3.0)))))) )))) - ((6.0)))))/2.0))  \n    v[\"i22\"] = 0.100000*np.tanh(((np.where((np.tanh((((data[\"pred_1\"]) + (-1.0))/2.0)) )<0, 2.0, (np.tanh((((data[\"pred_1\"]) < (-1.0))*1.)) ) )) * (((-3.0) * ((14.82431602478027344))))))  \n    v[\"i23\"] = 0.100000*np.tanh((((13.51132678985595703)) * ((np.tanh((((-3.0) + (data[\"t_sum_log\"]))/2.0)) ))))  \n    v[\"i24\"] = 0.100000*np.tanh((-1.0*((((np.minimum(((3.0)), (((np.tanh((((data[\"t_sum_log\"]) < ((9.0)))*1.)) ))))) * ((11.02354240417480469)))))))  \n    v[\"i25\"] = 0.100000*np.tanh((((((14.53642463684082031)) * (((np.where((14.62061119079589844)<0, (np.tanh(np.minimum(((data[\"t_sum_log\"])), ((3.0)))) ), (np.tanh(np.minimum(((data[\"t_sum_log\"])), ((3.0)))) ) )) - ((np.tanh((((3.0) < (data[\"pred_2\"]))*1.)) )))))) - ((14.53642177581787109))))  \n    v[\"i26\"] = 0.100000*np.tanh((((np.tanh(((data[\"t_sum_log\"]) + (-3.0))) )) * ((10.0))))  \n    v[\"i27\"] = 0.100000*np.tanh((((10.0)) * ((np.tanh((((-1.0) + ((((data[\"t_median\"]) > ((5.0)))*1.)))/2.0)) ))))  \n    v[\"i28\"] = 0.100000*np.tanh(((np.where((np.tanh(((-2.0) + (data[\"t_sum_log\"]))) )>0, 0.0, (-1.0*(((7.0)))) )) * 2.0))  \n    v[\"i29\"] = 0.100000*np.tanh(((((np.where(-3.0 < -99998, (np.tanh(((data[\"t_sum_act\"]) + ((-1.0*(((((((data[\"t_median\"]) + (-2.0))/2.0)) * (np.maximum(((data[\"pred_2\"])), ((-2.0))))))))))) ), ((((-2.0) + ((((np.tanh(((data[\"t_sum_act\"]) + ((-1.0*(((((((data[\"t_median\"]) + (-2.0))/2.0)) * (np.maximum(((data[\"pred_2\"])), ((-2.0))))))))))) )) * 2.0)))) * 2.0) )) * 2.0)) * 2.0))  \n    v[\"i30\"] = 0.100000*np.tanh(((np.where((8.40624523162841797)>0, (((np.tanh(np.minimum(((data[\"t_mean\"])), ((((((((data[\"t_sum_act\"]) + (-2.0))) * (data[\"t_sum_act\"]))) + (-1.0)))))) )) * ((8.40624523162841797))), (((np.tanh(np.minimum(((data[\"t_mean\"])), ((((((((data[\"t_sum_act\"]) + (-2.0))) * (data[\"t_sum_act\"]))) + (-1.0)))))) )) * ((8.40624141693115234))) )) - ((8.31408786773681641))))  \n    v[\"i31\"] = 0.100000*np.tanh((((((3.0)) * ((((np.where(3.0 < -99998, (9.90981674194335938), (np.tanh(((((data[\"t_sum_log\"]) - ((((np.maximum(((data[\"t_mean\"])), (((((data[\"t_mean\"]) > (data[\"t_mean\"]))*1.))))) + (data[\"visitNumber\"]))/2.0)))) * 2.0)) ) )) + (-1.0))/2.0)))) * ((9.90981674194335938))))  \n    v[\"i32\"] = 0.100000*np.tanh((((((((np.tanh(np.minimum(((1.0)), ((np.minimum(((((((((data[\"t_sum_log\"]) * ((1.95805239677429199)))) + (-2.0))) - ((2.0))))), ((data[\"geoNetwork_continent\"]))))))) )) * 2.0)) - ((2.54177641868591309)))) + ((((np.tanh(np.minimum(((1.0)), ((np.minimum(((((((((data[\"t_sum_log\"]) * ((1.95805239677429199)))) + (-2.0))) - ((2.0))))), ((data[\"geoNetwork_continent\"]))))))) )) * ((2.54177641868591309))))))  \n    v[\"i33\"] = 0.100000*np.tanh((((((13.79329872131347656)) * ((((np.tanh((((data[\"pred_0\"]) + (((((((data[\"t_sum_log\"]) - (((data[\"t_median\"]) * (data[\"visitNumber\"]))))) + (data[\"t_sum_log\"]))) - (data[\"visitNumber\"]))))/2.0)) )) + (np.where((np.tanh((((data[\"pred_0\"]) + (((((((data[\"t_sum_log\"]) - (((data[\"t_median\"]) * (data[\"visitNumber\"]))))) + (data[\"t_sum_log\"]))) - (data[\"visitNumber\"]))))/2.0)) ) < -99998, (np.tanh(np.minimum(((data[\"pred_3\"])), ((((((data[\"totals_pageviews\"]) + ((((np.maximum(((data[\"totals_pageviews\"])), ((-2.0)))) > (-2.0))*1.)))) + (data[\"t_mean\"])))))) ), -1.0 )))))) * 2.0))  \n    v[\"i34\"] = 0.100000*np.tanh((-1.0*((((((-1.0) + ((np.tanh(((data[\"t_sum_act\"]) + (-3.0))) )))) * ((-1.0*(((((((np.tanh(((data[\"t_sum_act\"]) + (-3.0))) )) + ((np.tanh(((data[\"t_sum_act\"]) + (-3.0))) )))) * (-3.0)))))))))))  \n    v[\"i35\"] = 0.100000*np.tanh(((np.minimum((((((np.tanh(((((data[\"pred_0\"]) * 2.0)) - (data[\"pred_3\"]))) )) * 2.0))), (((((((np.tanh(((((-1.0) + (((((data[\"t_sum_log\"]) + (-2.0))) + (data[\"t_sum_log\"]))))) * 2.0)) )) * 2.0)) - ((3.0))))))) + ((np.tanh(((((-1.0) + (((((data[\"t_sum_log\"]) + (-2.0))) + (data[\"t_sum_log\"]))))) * 2.0)) ))))  \n    v[\"i36\"] = 0.100000*np.tanh(((np.where(-1.0 < -99998, (np.tanh(((((((((((data[\"t_sum_act\"]) * (data[\"t_sum_act\"]))) + (-2.0))) * 2.0)) + (-1.0))) * 2.0)) ), (((((np.tanh(((((((((((data[\"t_sum_act\"]) * (data[\"t_sum_act\"]))) + (-2.0))) * 2.0)) + (-1.0))) * 2.0)) )) * 2.0)) + (-2.0)) )) * 2.0))  \n    v[\"i37\"] = 0.100000*np.tanh((-1.0*((np.where((np.tanh(((((data[\"t_mean\"]) * (np.where(data[\"pred_4\"]<0, data[\"t_sum_log\"], data[\"t_sum_log\"] )))) + (-3.0))) )>0, (((((np.tanh(np.maximum((((((data[\"t_nb_sess\"]) + (data[\"t_sum_act\"]))/2.0))), ((data[\"t_nb_sess\"])))) )) - (1.0))) * ((7.0))), (8.39564132690429688) )))))  \n    v[\"i38\"] = 0.100000*np.tanh(((((((np.where((np.tanh(((data[\"t_sum_log\"]) * (np.where(((data[\"t_sum_log\"]) - (np.minimum(((data[\"visitNumber\"])), ((data[\"t_sum_log\"])))))>0, data[\"t_sum_log\"], -1.0 )))) ) < -99998, (((np.tanh(((data[\"t_sum_log\"]) * (np.where(((data[\"t_sum_log\"]) - (np.minimum(((data[\"visitNumber\"])), ((data[\"t_sum_log\"])))))>0, data[\"t_sum_log\"], -1.0 )))) )) * 2.0), (((np.tanh(((data[\"t_sum_log\"]) * (np.where(((data[\"t_sum_log\"]) - (np.minimum(((data[\"visitNumber\"])), ((data[\"t_sum_log\"])))))>0, data[\"t_sum_log\"], -1.0 )))) )) * 2.0) )) - (2.0))) * 2.0)) * 2.0))  \n    v[\"i39\"] = 0.100000*np.tanh(((np.minimum(((((((((((-1.0) + ((np.tanh(((np.where(((data[\"pred_2\"]) + (data[\"t_sum_act\"])) < -99998, data[\"pred_5\"], data[\"t_sum_act\"] )) + (((data[\"t_sum_act\"]) + (np.tanh((-3.0))))))) )))) + ((np.tanh((((data[\"pred_6\"]) < (((((data[\"pred_6\"]) * 2.0)) + (data[\"t_mean\"]))))*1.)) )))) * 2.0)) * 2.0))), (((np.tanh(((np.where(((data[\"pred_2\"]) + (data[\"t_sum_act\"])) < -99998, data[\"pred_5\"], data[\"t_sum_act\"] )) + (((data[\"t_sum_act\"]) + (np.tanh((-3.0))))))) ))))) * 2.0))  \n    v[\"i40\"] = 0.100000*np.tanh(((((-2.0) + (((np.where(-2.0<0, np.where((np.tanh(((((((data[\"t_sum_act\"]) - (data[\"visitNumber\"]))) * 2.0)) + (-1.0))) )>0, (np.tanh(((((((data[\"t_sum_act\"]) - (data[\"visitNumber\"]))) * 2.0)) + (-1.0))) ), -2.0 ), -2.0 )) * 2.0)))) * 2.0))  \n    v[\"i41\"] = 0.100000*np.tanh(np.minimum(((((((((np.tanh(((((data[\"t_sum_log\"]) * ((-1.0*((data[\"totals_hits\"])))))) + (((((data[\"t_mean\"]) + (data[\"t_sum_log\"]))) + (data[\"t_sum_log\"]))))) )) - ((3.08904719352722168)))) + (2.0))/2.0))), ((((((3.0) * ((np.tanh(np.where(data[\"t_sum_act\"] < -99998, np.minimum(((((data[\"t_mean\"]) * 2.0))), ((np.maximum(((0.0)), ((data[\"t_sum_log\"])))))), data[\"t_sum_log\"] )) )))) - (3.0))))))  \n    v[\"i42\"] = 0.100000*np.tanh((((((-3.0) + ((((-2.0) + ((np.tanh(((((((((((data[\"t_sum_log\"]) * (data[\"t_sum_act\"]))) * (data[\"geoNetwork_continent\"]))) * (data[\"t_sum_act\"]))) * (data[\"t_sum_act\"]))) - (1.0))) )))/2.0)))/2.0)) + ((((np.tanh(((((((((((data[\"t_sum_log\"]) * (data[\"t_sum_act\"]))) * (data[\"geoNetwork_continent\"]))) * (data[\"t_sum_act\"]))) * (data[\"t_sum_act\"]))) - (1.0))) )) * 2.0))))  \n    v[\"i43\"] = 0.100000*np.tanh((((((((np.tanh(((((data[\"t_sum_log\"]) * (((data[\"t_sum_log\"]) * (((np.minimum(((data[\"t_sum_log\"])), ((data[\"t_sum_log\"])))) - (data[\"visitNumber\"]))))))) + (data[\"t_sum_log\"]))) )) + (-1.0))) * 2.0)) * 2.0))  \n    v[\"i44\"] = 0.100000*np.tanh(np.where((np.tanh(np.where(np.where(data[\"pred_1\"]>0, data[\"pred_2\"], data[\"t_sum_log\"] ) < -99998, data[\"pred_2\"], ((-1.0) + (data[\"t_sum_act\"])) )) )<0, ((-2.0) * ((8.40580558776855469))), (((np.tanh(((((np.tanh(np.where(np.where(data[\"pred_1\"]>0, data[\"pred_2\"], data[\"t_sum_log\"] ) < -99998, data[\"pred_2\"], ((-1.0) + (data[\"t_sum_act\"])) )) )) * ((4.0)))))) > ((np.tanh(np.where(np.where(data[\"pred_1\"]>0, data[\"pred_2\"], data[\"t_sum_log\"] ) < -99998, data[\"pred_2\"], ((-1.0) + (data[\"t_sum_act\"])) )) )))*1.) ))  \n    v[\"i45\"] = 0.100000*np.tanh(((((((((-1.0) + ((np.tanh(np.where(((((((data[\"t_sum_log\"]) * 2.0)) - (data[\"visitNumber\"]))) - (data[\"visitNumber\"]))>0, ((data[\"t_sum_log\"]) * 2.0), data[\"pred_3\"] )) )))) * 2.0)) * 2.0)) * 2.0))  \n    v[\"i46\"] = 0.100000*np.tanh(np.where((((np.tanh(np.where(data[\"visitNumber\"]<0, np.where(data[\"t_mean\"]>0, (((data[\"geoNetwork_continent\"]) + (data[\"visitNumber\"]))/2.0), data[\"pred_1\"] ), (((data[\"pred_1\"]) + (data[\"t_sum_log\"]))/2.0) )) )) / 2.0)>0, (((((np.tanh(np.where(data[\"visitNumber\"]<0, np.where(data[\"t_mean\"]>0, (((data[\"geoNetwork_continent\"]) + (data[\"visitNumber\"]))/2.0), data[\"pred_1\"] ), (((data[\"pred_1\"]) + (data[\"t_sum_log\"]))/2.0) )) )) / 2.0)) / 2.0), np.minimum(((((((-3.0) * 2.0)) * 2.0))), (((np.tanh(np.where(data[\"visitNumber\"]<0, np.where(data[\"t_mean\"]>0, (((data[\"geoNetwork_continent\"]) + (data[\"visitNumber\"]))/2.0), data[\"pred_1\"] ), (((data[\"pred_1\"]) + (data[\"t_sum_log\"]))/2.0) )) )))) ))  \n    v[\"i47\"] = 0.100000*np.tanh((((7.32015085220336914)) * (((np.tanh((1.0))) * ((((np.tanh(((((((np.minimum(((data[\"t_sum_log\"])), ((data[\"geoNetwork_metro\"])))) * 2.0)) + (np.where(data[\"pred_0\"]>0, data[\"pred_0\"], data[\"t_sum_log\"] )))) * 2.0)) )) + (-1.0)))))))  \n    v[\"i48\"] = 0.100000*np.tanh((((((((np.tanh(((np.tanh((data[\"t_sum_log\"]))) + ((((data[\"t_sum_log\"]) + (data[\"t_sum_log\"]))/2.0)))) )) - (((((((0.0) < ((np.tanh(((np.tanh((data[\"t_sum_log\"]))) + ((((data[\"t_sum_log\"]) + (data[\"t_sum_log\"]))/2.0)))) )))*1.)) > (((((np.tanh(((np.tanh((data[\"t_sum_log\"]))) + ((((data[\"t_sum_log\"]) + (data[\"t_sum_log\"]))/2.0)))) )) + ((np.tanh(((np.tanh((data[\"t_sum_log\"]))) + ((((data[\"t_sum_log\"]) + (data[\"t_sum_log\"]))/2.0)))) )))/2.0)))*1.)))) * 2.0)) * 2.0))  \n    v[\"i49\"] = 0.100000*np.tanh((((((np.where((np.tanh(np.where(np.where(((data[\"t_sum_act\"]) + (data[\"t_sum_log\"]))<0, data[\"pred_4\"], (((data[\"totals_pageviews\"]) > (data[\"totals_hits\"]))*1.) )>0, data[\"totals_hits\"], data[\"pred_4\"] )) )>0, np.where((np.tanh(np.where(np.where(((data[\"t_sum_act\"]) + (data[\"t_sum_log\"]))<0, data[\"pred_4\"], (((data[\"totals_pageviews\"]) > (data[\"totals_hits\"]))*1.) )>0, data[\"totals_hits\"], data[\"pred_4\"] )) )>0, (np.tanh(np.where(np.where(((data[\"t_sum_act\"]) + (data[\"t_sum_log\"]))<0, data[\"pred_4\"], (((data[\"totals_pageviews\"]) > (data[\"totals_hits\"]))*1.) )>0, data[\"totals_hits\"], data[\"pred_4\"] )) ), -3.0 ), -3.0 )) + (np.tanh((-1.0))))/2.0)) * 2.0))  \n    v[\"i50\"] = 0.100000*np.tanh(np.minimum((((((np.tanh(((((data[\"t_sum_act\"]) / 2.0)) + (((data[\"t_mean\"]) - (((1.0) * 2.0)))))) )) * ((((11.44591140747070312)) + (1.0)))))), (((((1.0) > ((((1.0) > ((np.tanh(((((data[\"t_sum_act\"]) / 2.0)) + (((data[\"t_mean\"]) - (((1.0) * 2.0)))))) )))*1.)))*1.)))))  \n    v[\"i51\"] = 0.100000*np.tanh(np.where((np.tanh((((data[\"t_mean\"]) + (np.where((((data[\"t_sum_log\"]) + (data[\"pred_1\"]))/2.0)<0, data[\"t_sum_log\"], np.tanh((data[\"t_mean\"])) )))/2.0)) )>0, ((((np.tanh((((data[\"t_mean\"]) + (np.where((((data[\"t_sum_log\"]) + (data[\"pred_1\"]))/2.0)<0, data[\"t_sum_log\"], np.tanh((data[\"t_mean\"])) )))/2.0)) )) < (((((np.tanh((((data[\"t_mean\"]) + (np.where((((data[\"t_sum_log\"]) + (data[\"pred_1\"]))/2.0)<0, data[\"t_sum_log\"], np.tanh((data[\"t_mean\"])) )))/2.0)) )) < (np.tanh(((3.0)))))*1.)))*1.), -3.0 ))  \n    v[\"i52\"] = 0.099756*np.tanh(np.where((np.tanh(np.where(np.minimum(((np.where(data[\"t_sum_act\"] < -99998, data[\"geoNetwork_subContinent\"], data[\"t_mean\"] ))), ((((data[\"geoNetwork_subContinent\"]) * 2.0))))<0, data[\"pred_3\"], np.tanh((data[\"t_sum_act\"])) )) )>0, ((np.maximum((((np.tanh(np.where(np.minimum(((np.where(data[\"t_sum_act\"] < -99998, data[\"geoNetwork_subContinent\"], data[\"t_mean\"] ))), ((((data[\"geoNetwork_subContinent\"]) * 2.0))))<0, data[\"pred_3\"], np.tanh((data[\"t_sum_act\"])) )) ))), (((np.tanh(np.where(np.minimum(((np.where(data[\"t_sum_act\"] < -99998, data[\"geoNetwork_subContinent\"], data[\"t_mean\"] ))), ((((data[\"geoNetwork_subContinent\"]) * 2.0))))<0, data[\"pred_3\"], np.tanh((data[\"t_sum_act\"])) )) ))))) - (np.tanh((2.0)))), ((-3.0) - (2.0)) ))  \n    v[\"i53\"] = 0.100000*np.tanh((((((((((((((np.tanh(((((((-1.0) + (((((((data[\"t_sum_log\"]) * ((9.82299137115478516)))) - (3.0))) * 2.0)))) - (1.0))) * 2.0)) )) + ((((np.tanh(((((((-1.0) + (((((((data[\"t_sum_log\"]) * ((9.82299137115478516)))) - (3.0))) * 2.0)))) - (1.0))) * 2.0)) )) - ((np.tanh((((data[\"totals_hits\"]) + ((((((data[\"t_mean\"]) + (data[\"t_sum_log\"]))/2.0)) + (data[\"geoNetwork_metro\"]))))/2.0)) )))))) - ((np.tanh(((((((-1.0) + (((((((data[\"t_sum_log\"]) * ((9.82299137115478516)))) - (3.0))) * 2.0)))) - (1.0))) * 2.0)) )))) * 2.0)) * 2.0)) * 2.0)) * 2.0))  \n    v[\"i54\"] = 0.100000*np.tanh((((((((np.tanh(np.where(data[\"pred_4\"]<0, ((data[\"t_median\"]) + (np.where(data[\"pred_1\"]<0, ((data[\"t_sum_act\"]) * 2.0), data[\"totals_pageviews\"] ))), data[\"totals_pageviews\"] )) )) + (-1.0))) * 2.0)) * 2.0))  \n    v[\"i55\"] = 0.100000*np.tanh(((((((((np.tanh(np.minimum(((np.minimum(((((data[\"geoNetwork_metro\"]) + (np.minimum(((((data[\"geoNetwork_metro\"]) + (1.0)))), ((data[\"t_sum_act\"]))))))), ((data[\"t_sum_act\"]))))), ((data[\"t_mean\"])))) )) < (0.0))*1.)) * (-3.0))) * 2.0))  \n    v[\"i56\"] = 0.099951*np.tanh(np.where((np.tanh(((np.tanh((data[\"pred_1\"]))) + ((((((data[\"t_sum_log\"]) + (((((data[\"t_median\"]) * 2.0)) * (((data[\"pred_0\"]) * 2.0)))))/2.0)) * 2.0)))) )<0, np.minimum(((np.minimum(((-3.0)), ((np.maximum(((0.0)), ((0.0)))))))), ((-3.0))), 0.0 ))  \n    v[\"i57\"] = 0.098095*np.tanh(((np.minimum((((((14.66910171508789062)) * ((((((14.66910171508789062)) * ((np.tanh((((-1.0) + (np.where(((data[\"totals_hits\"]) + (-3.0))>0, data[\"geoNetwork_metro\"], ((data[\"t_mean\"]) + (((data[\"t_sum_act\"]) * 2.0))) )))/2.0)) )))) * 2.0))))), ((np.minimum(((0.0)), ((0.0))))))) * 2.0))  \n    v[\"i58\"] = 0.100000*np.tanh(np.where((np.tanh(((((((np.minimum(((((data[\"totals_hits\"]) - (data[\"totals_pageviews\"])))), ((data[\"t_mean\"])))) * 2.0)) * 2.0)) + (np.tanh((data[\"totals_hits\"]))))) )>0, np.where((0.24389153718948364)>0, np.where((np.tanh(((((((np.minimum(((((data[\"totals_hits\"]) - (data[\"totals_pageviews\"])))), ((data[\"t_mean\"])))) * 2.0)) * 2.0)) + (np.tanh((data[\"totals_hits\"]))))) )>0, (0.24389153718948364), -3.0 ), (0.24389153718948364) ), -3.0 ))  \n    v[\"i59\"] = 0.051295*np.tanh(np.minimum((((np.tanh(np.where(np.where(((data[\"t_mean\"]) * 2.0)>0, data[\"t_mean\"], data[\"t_mean\"] )>0, (((data[\"t_mean\"]) > (data[\"t_sum_log\"]))*1.), -3.0 )) ))), ((((3.0) * (np.where((np.tanh(np.where(np.where(((data[\"t_mean\"]) * 2.0)>0, data[\"t_mean\"], data[\"t_mean\"] )>0, (((data[\"t_mean\"]) > (data[\"t_sum_log\"]))*1.), -3.0 )) )<0, -3.0, (np.tanh(np.where(np.where(((data[\"t_mean\"]) * 2.0)>0, data[\"t_mean\"], data[\"t_mean\"] )>0, (((data[\"t_mean\"]) > (data[\"t_sum_log\"]))*1.), -3.0 )) ) )))))))  \n    v[\"i60\"] = 0.099853*np.tanh(((((-1.0) + ((np.tanh(((np.where(data[\"t_sum_log\"]>0, (((np.where(data[\"pred_1\"]<0, data[\"t_sum_act\"], 2.0 )) > (data[\"pred_3\"]))*1.), data[\"pred_3\"] )) * 2.0)) )))) * 2.0))  \n    v[\"i61\"] = 0.099805*np.tanh((((3.0)) * (np.minimum((((np.tanh(((data[\"t_sum_log\"]) + (np.minimum(((np.minimum(((((data[\"t_sum_log\"]) - (data[\"visitNumber\"])))), ((((data[\"t_mean\"]) + (data[\"t_median\"]))))))), ((data[\"t_median\"])))))) ))), ((((3.0) * (np.minimum(((0.0)), (((np.tanh(((data[\"t_sum_log\"]) + (np.minimum(((np.minimum(((((data[\"t_sum_log\"]) - (data[\"visitNumber\"])))), ((((data[\"t_mean\"]) + (data[\"t_median\"]))))))), ((data[\"t_median\"])))))) ))))))))))))  \n    v[\"i62\"] = 0.083781*np.tanh((((((np.tanh(((data[\"totals_hits\"]) + (((data[\"t_median\"]) * (np.maximum(((np.minimum(((data[\"t_sum_act\"])), ((data[\"t_sum_log\"]))))), ((((data[\"t_sum_act\"]) + (data[\"t_median\"])))))))))) )) - (np.tanh((((2.0) + (2.0))))))) * (2.0)))  \n    v[\"i63\"] = 0.099951*np.tanh(np.where((np.tanh(np.minimum((((((data[\"t_mean\"]) < (data[\"t_sum_act\"]))*1.))), ((((((data[\"t_mean\"]) + (data[\"t_sum_act\"]))) + (data[\"t_sum_act\"])))))) )<0, ((-3.0) * 2.0), np.minimum((((((np.tanh(np.minimum((((((data[\"t_mean\"]) < (data[\"t_sum_act\"]))*1.))), ((((((data[\"t_mean\"]) + (data[\"t_sum_act\"]))) + (data[\"t_sum_act\"])))))) )) * 2.0))), ((np.tanh((np.minimum((((np.tanh(np.minimum((((((data[\"t_mean\"]) < (data[\"t_sum_act\"]))*1.))), ((((((data[\"t_mean\"]) + (data[\"t_sum_act\"]))) + (data[\"t_sum_act\"])))))) ))), (((np.tanh(np.minimum((((((data[\"t_mean\"]) < (data[\"t_sum_act\"]))*1.))), ((((((data[\"t_mean\"]) + (data[\"t_sum_act\"]))) + (data[\"t_sum_act\"])))))) ))))))))) ))  \n    v[\"i64\"] = 0.076014*np.tanh(np.minimum(((np.minimum((((np.tanh(((((-2.0) + (((data[\"t_sum_log\"]) / 2.0)))) * ((12.65624237060546875)))) ))), (((((np.tanh(((((-2.0) + (((data[\"t_sum_log\"]) / 2.0)))) * ((12.65624237060546875)))) )) + (-1.0))))))), (((np.tanh(((((-2.0) + (((data[\"t_sum_log\"]) / 2.0)))) * ((12.65624237060546875)))) )))))  \n    v[\"i65\"] = 0.046165*np.tanh(np.minimum(((np.minimum((((np.tanh(((data[\"t_sum_act\"]) - (data[\"t_sum_log\"]))) ))), (((((0.0) > ((np.tanh(((data[\"t_sum_act\"]) - (data[\"t_sum_log\"]))) )))*1.)))))), ((0.0))))  \n    v[\"i66\"] = 0.100000*np.tanh(np.minimum((((((((np.tanh(((((data[\"t_sum_log\"]) * 2.0)) + ((((np.where(np.minimum(((data[\"t_sum_log\"])), ((data[\"visitNumber\"])))<0, data[\"totals_hits\"], data[\"pred_1\"] )) + (-2.0))/2.0)))) )) * 2.0)) * 2.0))), ((((np.minimum((((0.06700874865055084))), (((np.tanh(((((data[\"t_sum_log\"]) * 2.0)) + ((((np.where(np.minimum(((data[\"t_sum_log\"])), ((data[\"visitNumber\"])))<0, data[\"totals_hits\"], data[\"pred_1\"] )) + (-2.0))/2.0)))) ))))) * 2.0)))))  \n    v[\"i67\"] = 0.098828*np.tanh(np.minimum((((((0.0) > ((((0.0) > (0.0))*1.)))*1.))), (((np.tanh(((data[\"t_sum_log\"]) * (2.0))) )))))  \n    v[\"i68\"] = 0.100000*np.tanh(((np.minimum((((((np.tanh(np.minimum(((((data[\"totals_pageviews\"]) - (((((data[\"totals_pageviews\"]) - (data[\"totals_hits\"]))) * ((10.0))))))), ((((data[\"totals_pageviews\"]) - (data[\"totals_hits\"])))))) )) * 2.0))), (((((np.tanh(np.minimum(((((data[\"totals_pageviews\"]) - (((((data[\"totals_pageviews\"]) - (data[\"totals_hits\"]))) * ((10.0))))))), ((((data[\"totals_pageviews\"]) - (data[\"totals_hits\"])))))) )) * 2.0))))) * 2.0))  \n    v[\"i69\"] = 0.049731*np.tanh(np.minimum((((0.04890204593539238))), ((((-2.0) * (np.maximum((((np.tanh((((((data[\"t_mean\"]) + (-1.0))) > (data[\"t_sum_log\"]))*1.)) ))), (((0.04890204593539238))))))))))  \n    v[\"i70\"] = 0.098925*np.tanh(np.minimum(((0.0)), (((np.tanh(((data[\"t_sum_log\"]) - (1.0))) )))))  \n    v[\"i71\"] = 0.068100*np.tanh(((np.tanh((((np.tanh((np.tanh((((-3.0) + (-2.0))))))) + ((np.tanh(np.minimum((((((((data[\"t_sum_act\"]) - (((2.0) * (np.tanh((-1.0))))))) < ((7.0)))*1.))), ((data[\"pred_2\"])))) )))))) + (1.0)))  \n    v[\"i72\"] = 0.100000*np.tanh(np.minimum(((((np.minimum((((np.tanh((((7.90549468994140625)) + (((((((data[\"t_sum_log\"]) * ((-1.0*((data[\"pred_2\"])))))) + (data[\"t_mean\"]))) + (data[\"t_mean\"]))))) ))), (((np.tanh((((7.90549468994140625)) + (((((((data[\"t_sum_log\"]) * ((-1.0*((data[\"pred_2\"])))))) + (data[\"t_mean\"]))) + (data[\"t_mean\"]))))) ))))) + ((np.tanh((((7.90549468994140625)) + (((((((data[\"t_sum_log\"]) * ((-1.0*((data[\"pred_2\"])))))) + (data[\"t_mean\"]))) + (data[\"t_mean\"]))))) ))))), ((((((np.tanh((((7.90549468994140625)) + (((((((data[\"t_sum_log\"]) * ((-1.0*((data[\"pred_2\"])))))) + (data[\"t_mean\"]))) + (data[\"t_mean\"]))))) )) < (((((np.tanh((((7.90549468994140625)) + (((((((data[\"t_sum_log\"]) * ((-1.0*((data[\"pred_2\"])))))) + (data[\"t_mean\"]))) + (data[\"t_mean\"]))))) )) > ((0.0)))*1.)))*1.)))))  \n    v[\"i73\"] = 0.100000*np.tanh((((0.0) < ((np.tanh((((np.maximum(((data[\"t_mean\"])), ((0.0)))) < (data[\"totals_pageviews\"]))*1.)) )))*1.))  \n    v[\"i74\"] = 0.099707*np.tanh(np.minimum(((((np.minimum((((np.tanh(np.minimum(((((((((((((data[\"geoNetwork_continent\"]) + (data[\"geoNetwork_metro\"]))) * ((6.0)))) + (data[\"geoNetwork_metro\"]))) * 2.0)) * 2.0))), ((data[\"t_median\"])))) ))), ((0.0)))) * 2.0))), ((((np.minimum((((np.tanh(np.minimum(((((((((((((data[\"geoNetwork_continent\"]) + (data[\"geoNetwork_metro\"]))) * ((6.0)))) + (data[\"geoNetwork_metro\"]))) * 2.0)) * 2.0))), ((data[\"t_median\"])))) ))), ((0.0)))) * 2.0)))))  \n    v[\"i75\"] = 0.099951*np.tanh(((((((np.tanh((((data[\"pred_2\"]) > (np.maximum((((5.28380632400512695))), ((data[\"pred_0\"])))))*1.)) )) * 2.0)) > (0.0))*1.))  \n    v[\"i76\"] = 0.099951*np.tanh(np.minimum((((np.tanh(((data[\"trafficSource_referralPath\"]) * (-2.0))) ))), ((0.0))))  \n    v[\"i77\"] = 0.100000*np.tanh(((((np.minimum(((np.minimum((((0.04203201457858086))), ((2.0))))), (((np.tanh(np.minimum(((data[\"totals_pageviews\"])), (((((((data[\"t_mean\"]) * ((-1.0*((data[\"trafficSource_referralPath\"])))))) + (data[\"visitNumber\"]))/2.0))))) ))))) * (2.0))) * 2.0))  \n    v[\"i78\"] = 0.099951*np.tanh((((np.minimum((((np.tanh(((data[\"t_nb_sess\"]) + ((4.0)))) ))), (((np.tanh(np.maximum(((2.0)), ((data[\"totals_hits\"])))) ))))) < (0.0))*1.))  \n    v[\"i79\"] = 0.099902*np.tanh(((((np.tanh((((((0.0) + (data[\"geoNetwork_continent\"]))/2.0)) * ((4.63319158554077148)))) )) < (0.0))*1.))  \n    v[\"i80\"] = 0.099756*np.tanh(((((((np.tanh(((data[\"pred_2\"]) * ((((data[\"pred_2\"]) < (3.0))*1.)))) )) + ((((((((((((((-1.0) + ((np.tanh(((data[\"pred_2\"]) * ((((data[\"pred_2\"]) < (3.0))*1.)))) )))/2.0)) / 2.0)) / 2.0)) / 2.0)) / 2.0)) / 2.0)))/2.0)) / 2.0))  \n    v[\"i81\"] = 0.099463*np.tanh(((((((0.0) < ((np.tanh((((data[\"totals_hits\"]) > ((4.0)))*1.)) )))*1.)) > ((np.tanh((((data[\"totals_hits\"]) > ((4.0)))*1.)) )))*1.))  \n    v[\"i82\"] = 0.099756*np.tanh((((0.0) < ((np.tanh((((3.0) > (data[\"t_sum_log\"]))*1.)) )))*1.))  \n    v[\"i83\"] = 0.099951*np.tanh(np.minimum(((((((np.tanh(np.minimum(((((((data[\"t_sum_act\"]) - ((((data[\"totals_hits\"]) + (data[\"totals_hits\"]))/2.0)))) + (data[\"visitNumber\"])))), ((((data[\"visitNumber\"]) - (data[\"trafficSource_referralPath\"])))))) )) < (np.minimum((((-1.0*(((((-1.0) < (((2.0) / 2.0)))*1.)))))), (((np.tanh(np.minimum(((((((data[\"t_sum_act\"]) - ((((data[\"totals_hits\"]) + (data[\"totals_hits\"]))/2.0)))) + (data[\"visitNumber\"])))), ((((data[\"visitNumber\"]) - (data[\"trafficSource_referralPath\"])))))) ))))))*1.))), (((np.tanh(np.minimum(((((((data[\"t_sum_act\"]) - ((((data[\"totals_hits\"]) + (data[\"totals_hits\"]))/2.0)))) + (data[\"visitNumber\"])))), ((((data[\"visitNumber\"]) - (data[\"trafficSource_referralPath\"])))))) )))))  \n    v[\"i84\"] = 0.100000*np.tanh(((np.minimum((((((((np.minimum((((9.0))), (((np.tanh(((((data[\"channelGrouping\"]) + (((data[\"geoNetwork_continent\"]) + (data[\"geoNetwork_metro\"]))))) * (data[\"geoNetwork_metro\"]))) ))))) > (2.0))*1.)) * ((np.tanh(((((data[\"channelGrouping\"]) + (((data[\"geoNetwork_continent\"]) + (data[\"geoNetwork_metro\"]))))) * (data[\"geoNetwork_metro\"]))) ))))), (((np.tanh(((((data[\"channelGrouping\"]) + (((data[\"geoNetwork_continent\"]) + (data[\"geoNetwork_metro\"]))))) * (data[\"geoNetwork_metro\"]))) ))))) * ((11.39656257629394531))))  \n    v[\"i85\"] = 0.099951*np.tanh(((((((np.minimum((((np.tanh(np.minimum(((((((((1.0) < (data[\"t_sum_log\"]))*1.)) > (data[\"geoNetwork_metro\"]))*1.))), ((np.minimum(((data[\"t_sum_log\"])), (((((data[\"t_mean\"]) + (data[\"totals_hits\"]))/2.0)))))))) ))), (((0.03772259503602982))))) * 2.0)) * 2.0)) * 2.0))  \n    v[\"i86\"] = 0.099365*np.tanh(((((np.tanh((((-2.0) + (data[\"t_mean\"]))/2.0)) )) < (0.0))*1.))  \n    v[\"i87\"] = 0.099853*np.tanh(np.minimum(((np.minimum(((0.0)), ((np.minimum(((0.0)), ((0.0)))))))), ((np.minimum((((np.tanh(((((-2.0) + (np.minimum((((((((13.26135158538818359)) * (data[\"geoNetwork_country\"]))) - (data[\"geoNetwork_country\"])))), ((data[\"totals_hits\"])))))) * (data[\"trafficSource_source\"]))) ))), ((0.0)))))))  \n    v[\"i88\"] = 0.099951*np.tanh(np.minimum((((((np.tanh(np.where(data[\"t_sum_act\"]<0, data[\"t_sum_log\"], np.tanh(((((np.maximum(((data[\"pred_1\"])), (((5.75660467147827148))))) > (data[\"t_sum_log\"]))*1.))) )) )) - ((0.17407540977001190))))), (((np.tanh(np.where(data[\"t_sum_act\"]<0, data[\"t_sum_log\"], np.tanh(((((np.maximum(((data[\"pred_1\"])), (((5.75660467147827148))))) > (data[\"t_sum_log\"]))*1.))) )) )))))  \n    v[\"i89\"] = 0.099902*np.tanh(((np.tanh(((4.0)))) + (np.tanh((((np.where((np.tanh(((data[\"t_sum_log\"]) - (((np.maximum(((-2.0)), ((data[\"trafficSource_source\"])))) + (data[\"pred_2\"]))))) )<0, (13.98869228363037109), -1.0 )) + (-2.0)))))))  \n    v[\"i90\"] = 0.100000*np.tanh((-1.0*(((((((np.minimum(((0.0)), (((np.tanh(((np.maximum(((-1.0)), ((((np.minimum(((data[\"t_sum_log\"])), ((data[\"t_sum_act\"])))) * (data[\"channelGrouping\"])))))) - (((data[\"t_sum_log\"]) * (data[\"channelGrouping\"]))))) ))))) * (2.0))) < ((0.0)))*1.)))))  \n    v[\"i91\"] = 0.066097*np.tanh(((((0.05042554065585136)) + ((((((((np.tanh(((((((data[\"t_median\"]) - ((((data[\"geoNetwork_metro\"]) + ((((data[\"t_mean\"]) + (data[\"pred_2\"]))/2.0)))/2.0)))) * (data[\"pred_1\"]))) * 2.0)) )) + ((((np.tanh(((((((data[\"t_median\"]) - ((((data[\"geoNetwork_metro\"]) + ((((data[\"t_mean\"]) + (data[\"pred_2\"]))/2.0)))/2.0)))) * (data[\"pred_1\"]))) * 2.0)) )) / 2.0)))/2.0)) + ((np.tanh(((((((data[\"t_median\"]) - ((((data[\"geoNetwork_metro\"]) + ((((data[\"t_mean\"]) + (data[\"pred_2\"]))/2.0)))/2.0)))) * (data[\"pred_1\"]))) * 2.0)) )))/2.0)))/2.0))  \n    v[\"i92\"] = 0.100000*np.tanh((((((np.tanh(((((12.78093624114990234)) < (((data[\"pred_0\"]) + (np.maximum(((data[\"channelGrouping\"])), ((np.maximum(((data[\"totals_hits\"])), ((data[\"channelGrouping\"]))))))))))*1.)) )) + ((((np.tanh(((((12.78093624114990234)) < (((data[\"pred_0\"]) + (np.maximum(((data[\"channelGrouping\"])), ((np.maximum(((data[\"totals_hits\"])), ((data[\"channelGrouping\"]))))))))))*1.)) )) / 2.0)))) / 2.0))  \n    v[\"i93\"] = 0.100000*np.tanh((((6.0)) * ((((6.0)) * (np.minimum(((((((6.0)) < (np.minimum(((0.0)), ((0.0)))))*1.))), (((np.tanh(np.where(np.maximum(((((data[\"geoNetwork_continent\"]) + (data[\"geoNetwork_metro\"])))), ((-1.0)))>0, np.minimum(((data[\"t_sum_act\"])), ((data[\"t_sum_log\"]))), data[\"pred_2\"] )) )))))))))  \n    v[\"i94\"] = 0.077088*np.tanh((((np.where((np.tanh(np.where(data[\"pred_4\"] < -99998, ((((1.0) + (data[\"pred_4\"]))) / 2.0), ((data[\"pred_2\"]) - (data[\"trafficSource_source\"])) )) )<0, (0.01666546240448952), (12.52340126037597656) )) + ((0.01666546240448952)))/2.0))  \n    v[\"i95\"] = 0.082413*np.tanh((((((np.tanh(((((data[\"channelGrouping\"]) - (data[\"visitNumber\"]))) * 2.0)) )) / 2.0)) / 2.0))  \n    v[\"i96\"] = 0.100000*np.tanh(np.maximum((((np.tanh(np.tanh((np.tanh((((((((data[\"geoNetwork_continent\"]) + (((data[\"t_nb_sess\"]) / 2.0)))/2.0)) < (-1.0))*1.)))))) ))), ((-3.0))))  \n    v[\"i97\"] = 0.099902*np.tanh(((((np.tanh((((4.26091861724853516)) - (data[\"totals_pageviews\"]))) )) < (0.0))*1.))  \n    v[\"i98\"] = 0.100000*np.tanh((((0.0) > ((np.tanh(((data[\"t_nb_sess\"]) + ((7.0)))) )))*1.))  \n    v[\"i99\"] = 0.046654*np.tanh(((np.minimum(((0.0)), (((np.tanh(((data[\"t_sum_act\"]) + (((np.minimum((((((3.0) > (data[\"geoNetwork_metro\"]))*1.))), ((data[\"totals_pageviews\"])))) - (((data[\"t_mean\"]) + (data[\"geoNetwork_subContinent\"])))))))))))) * 2.0))\n\n    return Output(v.sum(axis=1))\n\ndef GP2(data):\n    v = pd.DataFrame()\n    v[\"i0\"] = 0.100000*np.tanh(((np.tanh((data[\"pred_4\"]))) + (((((((((data[\"t_sum_act\"]) - (((3.0) * 2.0)))) * 2.0)) + (-2.0))) * 2.0)))) \n    v[\"i1\"] = 0.100000*np.tanh(((((((data[\"pred_4\"]) + (((((np.maximum(((data[\"t_median\"])), ((data[\"pred_3\"])))) - ((4.92847061157226562)))) + (data[\"pred_2\"]))))) * 2.0)) * 2.0)) \n    v[\"i2\"] = 0.100000*np.tanh(((((-1.0) - (data[\"t_sum_log\"]))) + (((((((data[\"t_sum_log\"]) - ((13.55201435089111328)))) + (data[\"t_mean\"]))) * ((13.55201435089111328)))))) \n    v[\"i3\"] = 0.100000*np.tanh(((3.0) * (((data[\"t_sum_act\"]) - (np.where(np.where(data[\"pred_2\"] < -99998, data[\"pred_1\"], 3.0 )>0, (7.0), (9.0) )))))) \n    v[\"i4\"] = 0.100000*np.tanh(((data[\"t_mean\"]) + ((((((((9.21076107025146484)) / 2.0)) - ((((9.21076107025146484)) * 2.0)))) + (np.maximum((((3.85717129707336426))), ((data[\"t_sum_log\"])))))))) \n    v[\"i5\"] = 0.100000*np.tanh(((((np.maximum(((((((data[\"pred_4\"]) * 2.0)) + (data[\"pred_0\"])))), ((data[\"pred_0\"])))) - ((((12.73920822143554688)) - (data[\"geoNetwork_metro\"]))))) * 2.0)) \n    v[\"i6\"] = 0.100000*np.tanh(((((np.minimum(((((((data[\"t_median\"]) * (((data[\"visitNumber\"]) * 2.0)))) + (data[\"visitNumber\"])))), ((data[\"pred_1\"])))) * 2.0)) - ((9.0)))) \n    v[\"i7\"] = 0.100000*np.tanh((((((6.33722066879272461)) * (((((((data[\"t_sum_log\"]) - ((6.33722066879272461)))) + (((-3.0) * 2.0)))) + (data[\"t_mean\"]))))) * 2.0)) \n    v[\"i8\"] = 0.100000*np.tanh(((((np.maximum(((data[\"pred_4\"])), ((data[\"pred_2\"])))) + (data[\"pred_6\"]))) - (((3.0) - (np.maximum(((data[\"pred_4\"])), ((data[\"pred_1\"])))))))) \n    v[\"i9\"] = 0.100000*np.tanh(((((((((-2.0) + (data[\"t_sum_act\"]))) - ((3.93038845062255859)))) * 2.0)) * 2.0)) \n    v[\"i10\"] = 0.100000*np.tanh(((((((np.maximum(((data[\"pred_10\"])), ((np.maximum(((data[\"pred_13\"])), ((np.maximum(((data[\"pred_5\"])), ((data[\"pred_3\"])))))))))) - (2.0))) * 2.0)) * 2.0)) \n    v[\"i11\"] = 0.100000*np.tanh(((((((((data[\"t_sum_log\"]) - ((((9.0)) - (((data[\"t_mean\"]) + (-2.0))))))) * 2.0)) * 2.0)) * 2.0)) \n    v[\"i12\"] = 0.100000*np.tanh(((((data[\"t_sum_log\"]) - (((3.0) * 2.0)))) - (((((((((data[\"visitNumber\"]) > (2.0))*1.)) * 2.0)) + (data[\"visitNumber\"]))/2.0)))) \n    v[\"i13\"] = 0.100000*np.tanh(((((((-2.0) + (np.where(data[\"pred_2\"]>0, np.where(data[\"pred_3\"]>0, data[\"pred_5\"], data[\"pred_2\"] ), data[\"pred_3\"] )))) * 2.0)) * 2.0)) \n    v[\"i14\"] = 0.100000*np.tanh(np.where(((data[\"t_sum_log\"]) + (((data[\"t_mean\"]) - ((9.50519275665283203)))))<0, ((-3.0) - ((9.50519275665283203))), 1.0 )) \n    v[\"i15\"] = 0.100000*np.tanh(((((((((((data[\"pred_6\"]) * 2.0)) * 2.0)) + (data[\"pred_1\"]))) + (np.minimum(((data[\"pred_7\"])), ((-3.0)))))) * 2.0)) \n    v[\"i16\"] = 0.100000*np.tanh(((data[\"t_sum_act\"]) + (((((((data[\"t_sum_act\"]) * (2.0))) - (np.maximum(((3.0)), ((data[\"visitNumber\"])))))) - ((11.07681083679199219)))))) \n    v[\"i17\"] = 0.100000*np.tanh(((-3.0) + (((((-2.0) + (data[\"t_mean\"]))) - (np.maximum(((0.0)), ((((data[\"visitNumber\"]) * (data[\"visitNumber\"])))))))))) \n    v[\"i18\"] = 0.100000*np.tanh(((np.minimum((((((3.0)) - (((data[\"t_mean\"]) - ((4.0))))))), ((((data[\"t_sum_log\"]) - ((4.0))))))) * 2.0)) \n    v[\"i19\"] = 0.100000*np.tanh(((((data[\"t_sum_act\"]) - (np.maximum(((data[\"visitNumber\"])), ((np.maximum(((data[\"t_mean\"])), (((((7.68621540069580078)) - (data[\"t_mean\"]))))))))))) * 2.0)) \n    v[\"i20\"] = 0.100000*np.tanh(((-3.0) + (((np.where(((data[\"t_sum_log\"]) - (data[\"visitNumber\"]))>0, data[\"t_sum_log\"], ((data[\"t_sum_log\"]) / 2.0) )) + (-2.0))))) \n    v[\"i21\"] = 0.100000*np.tanh((-1.0*((np.maximum((((((data[\"t_sum_act\"]) > ((7.0)))*1.))), ((np.where((((data[\"t_sum_log\"]) + (-3.0))/2.0)<0, (7.0), data[\"pred_6\"] )))))))) \n    v[\"i22\"] = 0.100000*np.tanh(np.minimum(((((((((((data[\"t_sum_log\"]) - ((4.0)))) * ((13.82894325256347656)))) * (3.0))) * ((4.0))))), ((0.0)))) \n    v[\"i23\"] = 0.100000*np.tanh(((((((np.minimum(((0.0)), ((((((((data[\"t_sum_act\"]) * 2.0)) + ((-1.0*(((7.07149600982666016))))))) * 2.0))))) * 2.0)) * 2.0)) * 2.0)) \n    v[\"i24\"] = 0.100000*np.tanh(np.minimum(((((((data[\"t_sum_log\"]) + (np.minimum(((data[\"t_sum_log\"])), ((data[\"t_mean\"])))))) - ((6.0))))), (((((data[\"t_sum_act\"]) < ((6.0)))*1.))))) \n    v[\"i25\"] = 0.100000*np.tanh(np.where((((-3.0) + (data[\"t_sum_act\"]))/2.0)>0, (((((data[\"t_sum_log\"]) - (3.0))) > (data[\"t_mean\"]))*1.), (-1.0*(((9.0)))) )) \n    v[\"i26\"] = 0.100000*np.tanh((((((((3.0) > (data[\"t_sum_act\"]))*1.)) * (((-3.0) * 2.0)))) - ((((data[\"t_sum_act\"]) < (((data[\"totals_hits\"]) * 2.0)))*1.)))) \n    v[\"i27\"] = 0.100000*np.tanh(((np.where(((data[\"t_sum_log\"]) + (-3.0))>0, np.where(data[\"pred_8\"]<0, (0.0), data[\"t_sum_log\"] ), ((-3.0) * 2.0) )) * 2.0)) \n    v[\"i28\"] = 0.100000*np.tanh(((((np.minimum(((data[\"t_median\"])), ((((((data[\"t_sum_act\"]) - (3.0))) * ((((3.0) > (data[\"t_median\"]))*1.))))))) * 2.0)) * 2.0)) \n    v[\"i29\"] = 0.100000*np.tanh(np.minimum((((((((((-3.0) + (data[\"t_mean\"]))) * 2.0)) < (3.0))*1.))), ((((((data[\"t_sum_log\"]) - (3.0))) * 2.0))))) \n    v[\"i30\"] = 0.100000*np.tanh(np.where(((((data[\"t_sum_log\"]) * 2.0)) - ((3.95256853103637695)))>0, (((data[\"t_nb_sess\"]) < (((-3.0) * 2.0)))*1.), ((-3.0) * 2.0) )) \n    v[\"i31\"] = 0.100000*np.tanh(np.minimum(((np.minimum(((0.0)), ((np.minimum(((data[\"pred_0\"])), ((data[\"t_median\"])))))))), ((((data[\"t_sum_log\"]) - ((((5.0)) - (data[\"t_sum_log\"])))))))) \n    v[\"i32\"] = 0.100000*np.tanh(((((data[\"t_mean\"]) - ((10.0)))) * (((((((data[\"t_median\"]) > (data[\"pred_5\"]))*1.)) > (((((data[\"t_mean\"]) / 2.0)) / 2.0)))*1.)))) \n    v[\"i33\"] = 0.100000*np.tanh(((((np.minimum((((((2.0) > (((data[\"t_sum_act\"]) + (-2.0))))*1.))), ((((data[\"t_sum_log\"]) + (-2.0)))))) * 2.0)) * 2.0)) \n    v[\"i34\"] = 0.100000*np.tanh(((((((((((data[\"t_sum_log\"]) * (data[\"t_sum_log\"]))) + (-3.0))) * ((((3.0) > (data[\"t_sum_log\"]))*1.)))) * 2.0)) * 2.0)) \n    v[\"i35\"] = 0.100000*np.tanh(np.where((((-3.0) + (data[\"t_sum_log\"]))/2.0)>0, 0.0, ((((data[\"t_sum_log\"]) - (3.0))) * 2.0) )) \n    v[\"i36\"] = 0.100000*np.tanh((((-1.0*(((8.26050090789794922))))) * ((((data[\"t_sum_act\"]) < (np.maximum(((((data[\"t_median\"]) - (data[\"totals_hits\"])))), (((2.0))))))*1.)))) \n    v[\"i37\"] = 0.100000*np.tanh(np.minimum(((((((((data[\"t_sum_log\"]) - (data[\"visitNumber\"]))) * 2.0)) - (((2.0) * 2.0))))), (((((data[\"totals_pageviews\"]) > ((4.61927986145019531)))*1.))))) \n    v[\"i38\"] = 0.100000*np.tanh(((np.where((((data[\"t_sum_log\"]) < (1.0))*1.)>0, -3.0, (((-3.0) < ((-1.0*((data[\"t_sum_log\"])))))*1.) )) * 2.0)) \n    v[\"i39\"] = 0.100000*np.tanh((((-1.0*(((6.07970380783081055))))) * ((((data[\"t_sum_log\"]) < (((((3.29153609275817871)) + (np.maximum(((data[\"visitNumber\"])), ((data[\"totals_hits\"])))))/2.0)))*1.)))) \n    v[\"i40\"] = 0.100000*np.tanh(np.where(((((data[\"t_sum_act\"]) / 2.0)) - (np.tanh((1.0))))<0, ((-3.0) * 2.0), (((data[\"t_median\"]) < (0.0))*1.) )) \n    v[\"i41\"] = 0.100000*np.tanh(((np.where(((((((data[\"pred_0\"]) + (-3.0))/2.0)) + (data[\"t_sum_act\"]))/2.0)<0, -3.0, np.minimum(((0.0)), ((data[\"t_median\"]))) )) * 2.0)) \n    v[\"i42\"] = 0.100000*np.tanh(np.where((((((data[\"t_sum_log\"]) < (1.0))*1.)) * (data[\"pred_2\"]))<0, data[\"pred_5\"], (-1.0*(((((data[\"pred_2\"]) > (data[\"totals_pageviews\"]))*1.)))) )) \n    v[\"i43\"] = 0.100000*np.tanh(((((((((((data[\"geoNetwork_continent\"]) + (data[\"t_sum_act\"]))/2.0)) < (np.maximum(((data[\"pred_5\"])), ((1.0)))))*1.)) * 2.0)) * (((data[\"pred_3\"]) * 2.0)))) \n    v[\"i44\"] = 0.100000*np.tanh((((-1.0*(((7.0))))) * (np.where(data[\"t_mean\"]<0, 2.0, ((((((1.0) > (data[\"t_sum_log\"]))*1.)) > (data[\"t_sum_act\"]))*1.) )))) \n    v[\"i45\"] = 0.100000*np.tanh(np.where(data[\"t_mean\"]<0, ((data[\"t_mean\"]) - ((6.90487623214721680))), (((((data[\"totals_pageviews\"]) > (data[\"totals_hits\"]))*1.)) / 2.0) )) \n    v[\"i46\"] = 0.100000*np.tanh(((data[\"pred_4\"]) * (((((((data[\"pred_4\"]) < (data[\"geoNetwork_continent\"]))*1.)) > (((data[\"geoNetwork_continent\"]) * ((((data[\"totals_hits\"]) + (1.0))/2.0)))))*1.)))) \n    v[\"i47\"] = 0.100000*np.tanh(((((np.where(((data[\"t_sum_act\"]) + (np.tanh((-3.0))))>0, ((((2.22855496406555176)) > (data[\"t_mean\"]))*1.), -2.0 )) * 2.0)) * 2.0)) \n    v[\"i48\"] = 0.100000*np.tanh(np.where(data[\"t_mean\"]>0, np.minimum(((0.0)), ((np.where(data[\"geoNetwork_subContinent\"]>0, (((data[\"geoNetwork_subContinent\"]) + (data[\"geoNetwork_metro\"]))/2.0), -3.0 )))), -3.0 )) \n    v[\"i49\"] = 0.099805*np.tanh(np.where(((np.minimum(((data[\"geoNetwork_continent\"])), ((((data[\"t_mean\"]) - (1.0)))))) * 2.0)>0, ((data[\"t_sum_act\"]) - (data[\"t_mean\"])), -3.0 )) \n    v[\"i50\"] = 0.099756*np.tanh(np.minimum((((-1.0*((data[\"pred_3\"]))))), ((((((-3.0) * 2.0)) * ((((1.0) > (((data[\"t_sum_act\"]) * 2.0)))*1.))))))) \n    v[\"i51\"] = 0.099951*np.tanh(((np.where(np.minimum(((data[\"geoNetwork_subContinent\"])), ((np.minimum(((data[\"geoNetwork_continent\"])), ((data[\"t_sum_act\"]))))))<0, -3.0, np.maximum(((data[\"t_nb_sess\"])), ((0.0))) )) * 2.0)) \n    v[\"i52\"] = 0.098339*np.tanh(np.where(data[\"t_mean\"]<0, -3.0, np.minimum((((((np.where(data[\"geoNetwork_metro\"]<0, (3.0), data[\"totals_pageviews\"] )) > (data[\"totals_pageviews\"]))*1.))), ((data[\"t_mean\"]))) )) \n    v[\"i53\"] = 0.100000*np.tanh(np.minimum(((np.where(data[\"pred_3\"]<0, data[\"t_mean\"], data[\"geoNetwork_metro\"] ))), ((np.where(data[\"t_mean\"]>0, (((data[\"visitNumber\"]) < (data[\"pred_3\"]))*1.), data[\"pred_3\"] ))))) \n    v[\"i54\"] = 0.100000*np.tanh(np.where(np.minimum(((np.where(data[\"trafficSource_referralPath\"]>0, data[\"pred_5\"], data[\"t_mean\"] ))), ((data[\"geoNetwork_country\"])))>0, 0.0, ((-3.0) + (-3.0)) )) \n    v[\"i55\"] = 0.099951*np.tanh((-1.0*(((((((data[\"t_sum_log\"]) < (((np.maximum(((data[\"t_sum_act\"])), ((data[\"geoNetwork_subContinent\"])))) * (data[\"geoNetwork_subContinent\"]))))*1.)) * ((7.03404140472412109))))))) \n    v[\"i56\"] = 0.100000*np.tanh(np.where(data[\"t_sum_log\"]<0, data[\"pred_3\"], np.where(data[\"t_nb_sess\"]<0, 0.0, (-1.0*(((((data[\"totals_pageviews\"]) < (data[\"totals_hits\"]))*1.)))) ) )) \n    v[\"i57\"] = 0.100000*np.tanh(((((np.where(data[\"t_mean\"]<0, data[\"pred_3\"], (((data[\"pred_2\"]) > (((data[\"t_nb_sess\"]) + (data[\"t_sum_log\"]))))*1.) )) * 2.0)) * 2.0)) \n    v[\"i58\"] = 0.100000*np.tanh(((((np.where(data[\"t_sum_log\"]<0, -2.0, (-1.0*(((((((data[\"t_sum_log\"]) + (data[\"totals_pageviews\"]))) < (data[\"visitNumber\"]))*1.)))) )) * 2.0)) * 2.0)) \n    v[\"i59\"] = 0.099805*np.tanh(np.where(data[\"t_mean\"]<0, -3.0, (((data[\"t_sum_act\"]) < ((((data[\"t_sum_log\"]) + ((((data[\"t_mean\"]) + ((4.0)))/2.0)))/2.0)))*1.) )) \n    v[\"i60\"] = 0.099707*np.tanh(((np.where(data[\"t_sum_log\"]>0, (-1.0*(((((data[\"geoNetwork_continent\"]) < (np.tanh((1.0))))*1.)))), -3.0 )) * 2.0)) \n    v[\"i61\"] = 0.100000*np.tanh(np.where(data[\"t_sum_log\"]>0, ((((data[\"trafficSource_referralPath\"]) * 2.0)) * (((data[\"device_browser\"]) - (data[\"trafficSource_referralPath\"])))), ((data[\"trafficSource_referralPath\"]) - ((7.0))) )) \n    v[\"i62\"] = 0.099756*np.tanh(np.where(data[\"t_sum_log\"]<0, -3.0, ((np.where(data[\"t_median\"]<0, data[\"geoNetwork_metro\"], (((-3.0) > (data[\"geoNetwork_metro\"]))*1.) )) / 2.0) )) \n    v[\"i63\"] = 0.100000*np.tanh((-1.0*((((((((np.maximum(((data[\"totals_pageviews\"])), ((((data[\"t_median\"]) / 2.0))))) < (data[\"totals_hits\"]))*1.)) > (np.tanh((data[\"t_mean\"]))))*1.))))) \n    v[\"i64\"] = 0.089253*np.tanh(np.where(data[\"t_sum_act\"]<0, -3.0, ((data[\"totals_hits\"]) * ((((data[\"t_median\"]) < (np.where(data[\"pred_0\"]>0, data[\"pred_2\"], 2.0 )))*1.))) )) \n    v[\"i65\"] = 0.100000*np.tanh(np.where(data[\"pred_0\"]<0, ((data[\"pred_0\"]) * 2.0), (-1.0*(((((3.0) > (((data[\"pred_0\"]) + (((data[\"geoNetwork_metro\"]) * 2.0)))))*1.)))) )) \n    v[\"i66\"] = 0.100000*np.tanh((-1.0*((np.where(data[\"channelGrouping\"]>0, np.where(data[\"geoNetwork_metro\"]>0, 0.0, ((data[\"visitNumber\"]) / 2.0) ), np.minimum(((0.0)), ((data[\"geoNetwork_metro\"]))) ))))) \n    v[\"i67\"] = 0.100000*np.tanh(np.where(data[\"t_mean\"]>0, np.where(data[\"geoNetwork_continent\"]>0, (((-1.0*(((((data[\"geoNetwork_subContinent\"]) > (data[\"geoNetwork_continent\"]))*1.))))) * 2.0), data[\"pred_1\"] ), -3.0 )) \n    v[\"i68\"] = 0.083537*np.tanh(np.where(data[\"t_mean\"]<0, -3.0, (((((((data[\"geoNetwork_continent\"]) * (data[\"pred_1\"]))) > (((data[\"t_mean\"]) / 2.0)))*1.)) * (data[\"pred_2\"])) )) \n    v[\"i69\"] = 0.071373*np.tanh(np.where(data[\"t_sum_act\"]<0, ((-3.0) * 2.0), (((-3.0) > (((0.0) - (((data[\"geoNetwork_metro\"]) * 2.0)))))*1.) )) \n    v[\"i70\"] = 0.099853*np.tanh(np.where(data[\"geoNetwork_continent\"]<0, data[\"totals_pageviews\"], (-1.0*(((((data[\"pred_2\"]) > (((np.where(data[\"geoNetwork_metro\"]<0, data[\"pred_0\"], data[\"t_median\"] )) / 2.0)))*1.)))) )) \n    v[\"i71\"] = 0.099951*np.tanh(np.minimum(((data[\"totals_pageviews\"])), ((((data[\"t_sum_act\"]) * ((((data[\"t_mean\"]) < (np.where(data[\"pred_3\"]<0, data[\"totals_pageviews\"], data[\"visitNumber\"] )))*1.))))))) \n    v[\"i72\"] = 0.099902*np.tanh(np.where(np.minimum(((data[\"t_mean\"])), ((data[\"t_sum_log\"])))<0, data[\"pred_4\"], (((-1.0*(((((data[\"geoNetwork_metro\"]) > (((data[\"t_mean\"]) / 2.0)))*1.))))) * 2.0) )) \n    v[\"i73\"] = 0.100000*np.tanh(((((np.minimum(((np.minimum(((data[\"t_sum_act\"])), (((((data[\"pred_0\"]) < (data[\"t_median\"]))*1.)))))), (((((data[\"totals_hits\"]) < (data[\"trafficSource_referralPath\"]))*1.))))) * 2.0)) * 2.0)) \n    v[\"i74\"] = 0.100000*np.tanh(np.where(np.where(data[\"trafficSource_referralPath\"]<0, data[\"geoNetwork_country\"], (((data[\"trafficSource_referralPath\"]) > (data[\"geoNetwork_country\"]))*1.) )>0, np.minimum(((0.0)), ((data[\"geoNetwork_country\"]))), data[\"pred_2\"] ))\n    return Output(v.sum(axis=1))\n\n\ndef GP3(data):\n    v = pd.DataFrame()\n    v[\"i0\"] = 0.100000*np.tanh(((((((((((np.maximum(((data[\"t_sum_log\"])), ((data[\"t_mean\"])))) - ((8.52710533142089844)))) * 2.0)) + (np.tanh((data[\"t_mean\"]))))) * 2.0)) * 2.0)) \n    v[\"i1\"] = 0.100000*np.tanh(((((((((data[\"t_sum_act\"]) + (((np.minimum(((data[\"pred_1\"])), ((-2.0)))) - ((5.0)))))) * 2.0)) * 2.0)) * 2.0)) \n    v[\"i2\"] = 0.100000*np.tanh((((9.12465190887451172)) * ((((9.12465190887451172)) * (((((data[\"totals_hits\"]) + (((((data[\"t_sum_log\"]) - ((9.12465190887451172)))) * 2.0)))) * 2.0)))))) \n    v[\"i3\"] = 0.100000*np.tanh(np.minimum(((((((data[\"t_sum_log\"]) - ((((5.99947023391723633)) - (((data[\"t_mean\"]) - ((8.0)))))))) * 2.0))), ((data[\"visitNumber\"])))) \n    v[\"i4\"] = 0.100000*np.tanh(((((np.minimum(((((data[\"t_sum_log\"]) - (np.maximum((((7.0))), ((data[\"visitNumber\"]))))))), ((data[\"visitNumber\"])))) * 2.0)) * 2.0)) \n    v[\"i5\"] = 0.100000*np.tanh(((((data[\"t_sum_act\"]) * 2.0)) - (np.maximum(((data[\"t_sum_act\"])), ((np.maximum((((((((data[\"t_sum_act\"]) * 2.0)) + ((13.88008403778076172)))/2.0))), (((13.80931282043457031)))))))))) \n    v[\"i6\"] = 0.100000*np.tanh(((np.where(((-3.0) + (data[\"pred_3\"]))>0, data[\"totals_hits\"], ((data[\"pred_0\"]) + (((-3.0) - ((8.0))))) )) * 2.0)) \n    v[\"i7\"] = 0.100000*np.tanh(((((data[\"pred_2\"]) + (((((data[\"pred_4\"]) + (-2.0))) - ((((data[\"pred_9\"]) < (data[\"pred_2\"]))*1.)))))) * 2.0)) \n    v[\"i8\"] = 0.100000*np.tanh(((np.where(((((data[\"t_mean\"]) + (-3.0))) + (-3.0))>0, (9.18304538726806641), data[\"pred_3\"] )) + ((-1.0*(((7.48107624053955078))))))) \n    v[\"i9\"] = 0.100000*np.tanh(((data[\"pred_2\"]) - ((((((((3.65108466148376465)) - (data[\"pred_1\"]))) * ((((3.65108466148376465)) - (data[\"pred_2\"]))))) - (data[\"pred_1\"]))))) \n    v[\"i10\"] = 0.100000*np.tanh((((((11.89422702789306641)) * (((((data[\"t_sum_log\"]) + (2.0))) - ((8.0)))))) + (np.maximum(((data[\"pred_0\"])), ((2.0)))))) \n    v[\"i11\"] = 0.100000*np.tanh(((np.where(data[\"pred_8\"]>0, data[\"pred_5\"], ((data[\"pred_6\"]) + (np.minimum(((data[\"pred_7\"])), ((((-3.0) + (data[\"pred_6\"]))))))) )) * 2.0)) \n    v[\"i12\"] = 0.100000*np.tanh(((data[\"t_sum_log\"]) - (np.maximum(((np.maximum((((6.0))), (((4.0)))))), ((((((data[\"t_sum_log\"]) - ((6.0)))) * ((6.0))))))))) \n    v[\"i13\"] = 0.100000*np.tanh(((data[\"t_sum_log\"]) - (np.where(((data[\"t_sum_log\"]) + (((data[\"t_sum_log\"]) - ((9.73407459259033203)))))<0, (11.21657562255859375), data[\"visitNumber\"] )))) \n    v[\"i14\"] = 0.100000*np.tanh(((((((((((((((data[\"pred_4\"]) * 2.0)) - (3.0))) + (data[\"pred_5\"]))) + (data[\"pred_5\"]))) * 2.0)) * 2.0)) * 2.0)) \n    v[\"i15\"] = 0.100000*np.tanh(((((data[\"t_sum_act\"]) - ((5.0)))) * ((((4.0)) - (((((data[\"t_sum_act\"]) - ((5.0)))) * 2.0)))))) \n    v[\"i16\"] = 0.100000*np.tanh(((((((((np.where(((-3.0) + (data[\"visitNumber\"]))<0, data[\"t_sum_act\"], data[\"pred_7\"] )) - ((4.87859916687011719)))) * 2.0)) * 2.0)) * 2.0)) \n    v[\"i17\"] = 0.100000*np.tanh(((((((((((((((((data[\"t_mean\"]) * 2.0)) - ((8.0)))) * 2.0)) * 2.0)) * 2.0)) * 2.0)) + (data[\"pred_4\"]))) * 2.0)) \n    v[\"i18\"] = 0.100000*np.tanh(((((((data[\"t_mean\"]) + (((((data[\"t_sum_log\"]) + (((((data[\"t_sum_log\"]) - ((7.0)))) * 2.0)))) * 2.0)))) * 2.0)) * 2.0)) \n    v[\"i19\"] = 0.100000*np.tanh(((((3.0) - (data[\"t_median\"]))) + (((((data[\"pred_6\"]) - (np.where(data[\"t_median\"]<0, 3.0, (5.0) )))) * 2.0)))) \n    v[\"i20\"] = 0.100000*np.tanh(((((np.where(((data[\"t_mean\"]) - ((6.0)))>0, (4.0), data[\"t_sum_log\"] )) - ((4.0)))) * ((8.0)))) \n    v[\"i21\"] = 0.100000*np.tanh(((-3.0) + (np.where(((data[\"t_median\"]) - ((4.61104393005371094)))<0, ((data[\"t_mean\"]) - (3.0)), 3.0 )))) \n    v[\"i22\"] = 0.100000*np.tanh(((np.where((((-3.0) + (data[\"t_sum_log\"]))/2.0)>0, (((-3.0) + ((7.15739059448242188)))/2.0), (-1.0*(((7.15739059448242188)))) )) - (data[\"visitNumber\"]))) \n    v[\"i23\"] = 0.100000*np.tanh(((data[\"t_sum_act\"]) - (((np.maximum(((data[\"totals_pageviews\"])), ((np.maximum(((np.maximum(((data[\"pred_2\"])), ((((data[\"visitNumber\"]) / 2.0)))))), ((2.0))))))) * 2.0)))) \n    v[\"i24\"] = 0.100000*np.tanh(np.where(((-3.0) + (data[\"t_sum_act\"]))<0, ((data[\"t_sum_act\"]) - ((9.0))), (((data[\"t_nb_sess\"]) < (-3.0))*1.) )) \n    v[\"i25\"] = 0.100000*np.tanh(((data[\"t_sum_act\"]) - (np.maximum(((np.where((((-3.0) + (data[\"t_sum_act\"]))/2.0)<0, (7.0), data[\"pred_3\"] ))), ((data[\"t_sum_act\"])))))) \n    v[\"i26\"] = 0.100000*np.tanh(((((((data[\"t_sum_log\"]) - (3.0))) * ((((((((data[\"t_mean\"]) - (3.0))) * 2.0)) < (3.0))*1.)))) * 2.0)) \n    v[\"i27\"] = 0.100000*np.tanh(np.minimum((((((data[\"t_sum_act\"]) < ((5.0)))*1.))), ((np.minimum(((((data[\"t_sum_act\"]) - ((4.0))))), (((((4.0)) - (data[\"pred_1\"]))))))))) \n    v[\"i28\"] = 0.100000*np.tanh(((((((data[\"t_sum_log\"]) - ((13.52637577056884766)))) * ((((data[\"t_mean\"]) < (3.0))*1.)))) * ((((data[\"t_sum_log\"]) < (3.0))*1.)))) \n    v[\"i29\"] = 0.100000*np.tanh((((((1.0) > (((-3.0) + (data[\"t_sum_act\"]))))*1.)) * (((2.0) * (((-3.0) + (data[\"t_sum_log\"]))))))) \n    v[\"i30\"] = 0.100000*np.tanh((((((data[\"t_sum_log\"]) < (((np.tanh((np.maximum(((data[\"totals_hits\"])), ((3.0)))))) * 2.0)))*1.)) * (((data[\"t_sum_act\"]) - ((7.80285549163818359)))))) \n    v[\"i31\"] = 0.100000*np.tanh(np.where(((-2.0) + (data[\"t_sum_log\"]))>0, (((((-2.0) + (data[\"t_sum_log\"]))) > (data[\"t_sum_act\"]))*1.), (-1.0*(((8.0)))) )) \n    v[\"i32\"] = 0.100000*np.tanh(((np.where(((((data[\"t_sum_act\"]) - (2.0))) * (data[\"pred_6\"]))>0, -3.0, (((data[\"pred_6\"]) > (-3.0))*1.) )) * 2.0)) \n    v[\"i33\"] = 0.100000*np.tanh((((((2.0) > (data[\"t_sum_log\"]))*1.)) * ((((((2.0) > (data[\"t_sum_act\"]))*1.)) * (((((-3.0) * 2.0)) * 2.0)))))) \n    v[\"i34\"] = 0.100000*np.tanh(np.minimum(((np.minimum(((((-3.0) + (np.where(data[\"t_mean\"]>0, data[\"t_mean\"], -3.0 ))))), ((0.0))))), ((data[\"t_median\"])))) \n    v[\"i35\"] = 0.100000*np.tanh(((np.minimum((((((data[\"t_sum_log\"]) < (3.0))*1.))), ((np.minimum(((((((data[\"t_sum_act\"]) + (-2.0))) * 2.0))), ((data[\"t_median\"]))))))) * 2.0)) \n    v[\"i36\"] = 0.100000*np.tanh(np.minimum(((0.0)), (((((data[\"geoNetwork_metro\"]) + (((3.0) * ((((((((-2.0) + (data[\"t_sum_log\"]))/2.0)) * 2.0)) * 2.0)))))/2.0))))) \n    v[\"i37\"] = 0.100000*np.tanh(np.where(data[\"t_mean\"]<0, (-1.0*(((14.43007755279541016)))), (((data[\"pred_3\"]) > (np.where(data[\"pred_2\"]<0, data[\"visitNumber\"], data[\"t_mean\"] )))*1.) )) \n    v[\"i38\"] = 0.100000*np.tanh(np.minimum((((((((data[\"t_sum_log\"]) + (-3.0))) < (data[\"pred_2\"]))*1.))), ((((data[\"t_sum_log\"]) + (((data[\"t_sum_log\"]) + (-3.0)))))))) \n    v[\"i39\"] = 0.100000*np.tanh(((np.minimum(((data[\"t_median\"])), ((((data[\"t_sum_act\"]) - (np.maximum(((np.maximum(((2.0)), ((data[\"totals_hits\"]))))), ((data[\"t_sum_act\"]))))))))) * 2.0)) \n    v[\"i40\"] = 0.100000*np.tanh(((np.where(((data[\"t_sum_act\"]) - (1.0))<0, -3.0, ((((((8.0)) - (data[\"t_sum_log\"]))) < (-3.0))*1.) )) * 2.0)) \n    v[\"i41\"] = 0.100000*np.tanh(np.where(np.where(data[\"geoNetwork_continent\"]<0, data[\"pred_4\"], ((data[\"t_sum_log\"]) + (((data[\"t_sum_log\"]) + (-3.0)))) )>0, 0.0, -3.0 )) \n    v[\"i42\"] = 0.100000*np.tanh(((((data[\"pred_5\"]) * 2.0)) * (np.where((((3.0) > (data[\"pred_4\"]))*1.)>0, (((data[\"t_sum_log\"]) < (1.0))*1.), data[\"pred_4\"] )))) \n    v[\"i43\"] = 0.100000*np.tanh((((((((np.tanh((((data[\"t_sum_log\"]) * (data[\"t_sum_log\"]))))) * (data[\"geoNetwork_continent\"]))) < (((1.0) / 2.0)))*1.)) * (data[\"pred_5\"]))) \n    v[\"i44\"] = 0.100000*np.tanh(((np.where((((-1.0) + (data[\"t_mean\"]))/2.0)>0, np.where(data[\"pred_5\"]<0, 0.0, data[\"geoNetwork_metro\"] ), data[\"pred_1\"] )) * ((10.0)))) \n    v[\"i45\"] = 0.100000*np.tanh(np.where(data[\"t_sum_log\"]<0, -3.0, (((-2.0) + (np.where(((data[\"totals_hits\"]) - (data[\"totals_pageviews\"]))<0, 3.0, -3.0 )))/2.0) )) \n    v[\"i46\"] = 0.100000*np.tanh(((np.where(((data[\"t_sum_act\"]) + (np.tanh((-3.0))))>0, (((data[\"totals_hits\"]) > ((4.0)))*1.), (-1.0*(((4.0)))) )) * 2.0)) \n    v[\"i47\"] = 0.100000*np.tanh(((-3.0) + (np.where(data[\"geoNetwork_subContinent\"]>0, np.where(data[\"t_mean\"]>0, 3.0, -3.0 ), -3.0 )))) \n    v[\"i48\"] = 0.100000*np.tanh(np.where(np.where(data[\"t_mean\"]<0, data[\"pred_2\"], data[\"t_sum_log\"] )<0, -3.0, (((np.maximum(((data[\"pred_3\"])), ((data[\"pred_2\"])))) > (data[\"t_median\"]))*1.) )) \n    v[\"i49\"] = 0.100000*np.tanh(((-3.0) * ((((data[\"t_sum_log\"]) < ((((data[\"visitNumber\"]) + (np.maximum(((data[\"t_mean\"])), ((((3.0) / 2.0))))))/2.0)))*1.)))) \n    v[\"i50\"] = 0.100000*np.tanh(np.where(((((data[\"t_sum_act\"]) * 2.0)) + (-1.0))<0, -3.0, (((3.0) > (((data[\"t_sum_act\"]) + (-1.0))))*1.) )) \n    v[\"i51\"] = 0.100000*np.tanh(np.where((((((data[\"t_sum_act\"]) * 2.0)) + (-1.0))/2.0)<0, ((-1.0) - ((10.0))), (((data[\"geoNetwork_continent\"]) + (-1.0))/2.0) )) \n    v[\"i52\"] = 0.100000*np.tanh(((np.where((((data[\"t_sum_act\"]) + (np.tanh((data[\"pred_1\"]))))/2.0)<0, -3.0, (((data[\"t_sum_act\"]) < (data[\"pred_1\"]))*1.) )) * 2.0)) \n    v[\"i53\"] = 0.100000*np.tanh(((np.minimum(((np.where(data[\"t_mean\"]<0, -3.0, (((data[\"t_mean\"]) < (data[\"t_sum_log\"]))*1.) ))), (((((data[\"t_mean\"]) < (data[\"totals_pageviews\"]))*1.))))) * 2.0)) \n    v[\"i54\"] = 0.100000*np.tanh(((np.minimum(((((((data[\"t_sum_act\"]) * 2.0)) + (-3.0)))), (((((data[\"t_sum_act\"]) < (((data[\"t_sum_log\"]) + (-3.0))))*1.))))) * 2.0)) \n    v[\"i55\"] = 0.100000*np.tanh((((((-1.0*((np.maximum(((data[\"trafficSource_referralPath\"])), (((((data[\"t_sum_log\"]) < ((-1.0*((((data[\"visitNumber\"]) * (data[\"trafficSource_referralPath\"])))))))*1.)))))))) * 2.0)) * 2.0)) \n    v[\"i56\"] = 0.100000*np.tanh((((9.97595596313476562)) * ((-1.0*(((((0.0) > (((data[\"t_sum_act\"]) + (np.where(data[\"visitNumber\"]>0, data[\"pred_1\"], data[\"visitNumber\"] )))))*1.))))))) \n    v[\"i57\"] = 0.100000*np.tanh(np.where(data[\"t_mean\"]<0, data[\"pred_3\"], (((((np.where(data[\"pred_2\"]<0, data[\"t_mean\"], ((data[\"pred_2\"]) * 2.0) )) < (data[\"t_sum_act\"]))*1.)) / 2.0) )) \n    v[\"i58\"] = 0.099951*np.tanh((-1.0*((((((((-1.0*((data[\"pred_4\"])))) + (data[\"t_sum_act\"]))/2.0)) * ((((data[\"t_sum_act\"]) < (((data[\"t_mean\"]) * (data[\"pred_5\"]))))*1.))))))) \n    v[\"i59\"] = 0.099853*np.tanh(((np.minimum(((np.where(data[\"t_mean\"]>0, (((data[\"t_sum_log\"]) < (2.0))*1.), -3.0 ))), (((-1.0*((data[\"trafficSource_referralPath\"]))))))) * 2.0)) \n    v[\"i60\"] = 0.099511*np.tanh((((((((((data[\"visitNumber\"]) - (-3.0))) / 2.0)) > (data[\"t_sum_log\"]))*1.)) * (-3.0))) \n    v[\"i61\"] = 0.100000*np.tanh(((np.where(data[\"t_sum_act\"]<0, -3.0, (((((data[\"geoNetwork_subContinent\"]) > (np.tanh((((data[\"geoNetwork_subContinent\"]) * 2.0)))))*1.)) * (-3.0)) )) * 2.0)) \n    v[\"i62\"] = 0.099951*np.tanh(np.where(data[\"t_mean\"]<0, -3.0, (((-1.0*((((((-1.0*((np.tanh((np.tanh((data[\"t_sum_act\"])))))))) > (data[\"geoNetwork_metro\"]))*1.))))) * 2.0) )) \n    v[\"i63\"] = 0.100000*np.tanh(np.where(((((data[\"totals_hits\"]) * 2.0)) - (((data[\"t_median\"]) / 2.0)))<0, ((data[\"pred_2\"]) * 2.0), (((data[\"t_sum_log\"]) > ((9.24615192413330078)))*1.) )) \n    v[\"i64\"] = 0.100000*np.tanh(((np.where(data[\"t_sum_log\"]<0, -3.0, np.minimum(((((data[\"totals_pageviews\"]) - (data[\"totals_hits\"])))), ((((data[\"totals_hits\"]) - (data[\"totals_pageviews\"]))))) )) * 2.0)) \n    v[\"i65\"] = 0.099756*np.tanh(np.minimum(((data[\"totals_pageviews\"])), ((np.minimum(((np.tanh((np.tanh((data[\"geoNetwork_subContinent\"])))))), (((((data[\"trafficSource_referralPath\"]) + ((((data[\"geoNetwork_continent\"]) > (data[\"geoNetwork_subContinent\"]))*1.)))/2.0)))))))) \n    v[\"i66\"] = 0.099951*np.tanh(((np.where(data[\"t_sum_act\"]<0, data[\"pred_1\"], (((-1.0*(((((data[\"geoNetwork_metro\"]) < ((-1.0*((np.tanh((data[\"geoNetwork_continent\"])))))))*1.))))) * 2.0) )) * 2.0)) \n    v[\"i67\"] = 0.099756*np.tanh(((data[\"visitNumber\"]) * ((-1.0*(((((data[\"channelGrouping\"]) > (np.minimum(((np.minimum(((data[\"trafficSource_referralPath\"])), ((0.0))))), ((((data[\"visitNumber\"]) * 2.0))))))*1.))))))) \n    v[\"i68\"] = 0.100000*np.tanh(np.where(data[\"t_sum_act\"]<0, ((((-1.0) * 2.0)) * 2.0), (((data[\"pred_4\"]) > ((((((data[\"geoNetwork_metro\"]) * 2.0)) + (data[\"pred_1\"]))/2.0)))*1.) )) \n    v[\"i69\"] = 0.099902*np.tanh(np.where(data[\"t_sum_log\"]<0, data[\"pred_3\"], (((((((data[\"t_mean\"]) < (data[\"totals_hits\"]))*1.)) * (data[\"totals_hits\"]))) * (data[\"visitNumber\"])) )) \n    v[\"i70\"] = 0.100000*np.tanh((((((data[\"t_nb_sess\"]) < ((-1.0*((((((((data[\"geoNetwork_metro\"]) > (data[\"geoNetwork_continent\"]))*1.)) + (data[\"t_sum_log\"]))/2.0))))))*1.)) * (data[\"t_sum_log\"]))) \n    v[\"i71\"] = 0.099902*np.tanh(np.minimum((((((((data[\"totals_pageviews\"]) < ((2.0)))*1.)) * (((data[\"t_median\"]) * 2.0))))), ((((((2.0)) < (((data[\"t_median\"]) * 2.0)))*1.))))) \n    v[\"i72\"] = 0.099805*np.tanh(np.minimum(((data[\"t_sum_act\"])), (((((((data[\"t_mean\"]) < ((2.48734521865844727)))*1.)) * (np.where(data[\"t_mean\"]<0, data[\"pred_3\"], ((data[\"totals_pageviews\"]) / 2.0) ))))))) \n    v[\"i73\"] = 0.099902*np.tanh((-1.0*((np.where(data[\"pred_3\"]<0, 0.0, np.maximum(((data[\"pred_2\"])), (((((((data[\"totals_pageviews\"]) > (data[\"pred_3\"]))*1.)) * (data[\"pred_3\"]))))) ))))) \n    v[\"i74\"] = 0.100000*np.tanh(((((((np.where(data[\"t_mean\"]>0, (((data[\"geoNetwork_continent\"]) < ((((data[\"totals_hits\"]) + (-3.0))/2.0)))*1.), -3.0 )) * 2.0)) * 2.0)) * 2.0)) \n    v[\"i75\"] = 0.100000*np.tanh(np.where(data[\"geoNetwork_subContinent\"]>0, (((((data[\"pred_0\"]) > (((data[\"totals_hits\"]) - (data[\"pred_3\"]))))*1.)) * ((-1.0*((data[\"trafficSource_referralPath\"]))))), data[\"pred_1\"] ))\n    return Output(v.sum(axis=1))\n\ndef GP4(data):\n    v = pd.DataFrame()\n    v[\"i0\"] = 0.100000*np.tanh((((((((((((data[\"t_sum_act\"]) + (data[\"t_sum_log\"]))/2.0)) - ((7.28741741180419922)))) * 2.0)) * 2.0)) * 2.0)) \n    v[\"i1\"] = 0.100000*np.tanh(((np.maximum(((-2.0)), ((data[\"pred_4\"])))) + (((((((((data[\"t_mean\"]) - (3.0))) - ((4.06806802749633789)))) * 2.0)) * 2.0)))) \n    v[\"i2\"] = 0.100000*np.tanh(((((((((((((((((((data[\"t_sum_log\"]) - ((8.19811630249023438)))) * 2.0)) * 2.0)) * 2.0)) * 2.0)) * 2.0)) * 2.0)) * 2.0)) * 2.0)) \n    v[\"i3\"] = 0.100000*np.tanh(((np.minimum(((((data[\"t_sum_log\"]) - (data[\"visitNumber\"])))), ((((data[\"t_sum_log\"]) - ((7.0))))))) + (((data[\"t_sum_act\"]) - ((7.0)))))) \n    v[\"i4\"] = 0.100000*np.tanh((((((((((((((((data[\"t_sum_act\"]) + ((-1.0*(((7.0))))))/2.0)) * 2.0)) * 2.0)) * 2.0)) * 2.0)) * 2.0)) - (-3.0))) \n    v[\"i5\"] = 0.100000*np.tanh(np.minimum(((((data[\"t_mean\"]) + (((data[\"t_sum_log\"]) - ((12.75276947021484375))))))), ((np.minimum(((data[\"t_median\"])), ((((data[\"t_mean\"]) + (data[\"pred_1\"]))))))))) \n    v[\"i6\"] = 0.100000*np.tanh(((((((((((-1.0) + (((((((data[\"t_sum_log\"]) - ((9.0)))) * 2.0)) * 2.0)))) * 2.0)) * 2.0)) * 2.0)) * 2.0)) \n    v[\"i7\"] = 0.100000*np.tanh(((((((((data[\"t_sum_log\"]) - ((7.04534244537353516)))) - (np.tanh(((((data[\"t_sum_log\"]) + (data[\"pred_9\"]))/2.0)))))) * 2.0)) * 2.0)) \n    v[\"i8\"] = 0.100000*np.tanh(((((((((((data[\"pred_1\"]) + (((data[\"pred_2\"]) + (np.minimum(((-3.0)), ((data[\"pred_6\"])))))))) * 2.0)) * 2.0)) * 2.0)) * 2.0)) \n    v[\"i9\"] = 0.100000*np.tanh((((((((((4.59890556335449219)) + (((((((((data[\"t_median\"]) - ((8.0)))) * 2.0)) * 2.0)) * 2.0)))) * 2.0)) * 2.0)) * 2.0)) \n    v[\"i10\"] = 0.100000*np.tanh(((((((np.maximum(((data[\"pred_3\"])), ((np.maximum(((data[\"pred_4\"])), ((((data[\"pred_1\"]) + (data[\"pred_2\"]))))))))) + (-3.0))) * 2.0)) * 2.0)) \n    v[\"i11\"] = 0.100000*np.tanh(((((((((((((((data[\"t_sum_log\"]) - ((((4.0)) * 2.0)))) * 2.0)) + (data[\"t_mean\"]))) * 2.0)) * 2.0)) * 2.0)) * 2.0)) \n    v[\"i12\"] = 0.100000*np.tanh(((((((data[\"pred_10\"]) + (((((data[\"t_mean\"]) - (((data[\"visitNumber\"]) / 2.0)))) * 2.0)))) + (data[\"pred_3\"]))) * 2.0)) \n    v[\"i13\"] = 0.100000*np.tanh(((((data[\"t_mean\"]) + (np.minimum(((((((((data[\"t_sum_log\"]) - ((6.0)))) * 2.0)) + (-3.0)))), ((-3.0)))))) * 2.0)) \n    v[\"i14\"] = 0.100000*np.tanh(((((data[\"t_sum_act\"]) + (((np.minimum(((-3.0)), (((((-1.0*((data[\"visitNumber\"])))) * (data[\"visitNumber\"])))))) + (-2.0))))) * 2.0)) \n    v[\"i15\"] = 0.100000*np.tanh(np.minimum(((((data[\"pred_3\"]) + (((data[\"t_mean\"]) - ((4.77416133880615234))))))), (((((5.0)) + ((-1.0*((data[\"visitNumber\"]))))))))) \n    v[\"i16\"] = 0.100000*np.tanh(np.minimum(((((((data[\"t_sum_act\"]) - ((4.75877285003662109)))) * 2.0))), (((((((((data[\"t_sum_act\"]) - ((4.75877285003662109)))) * 2.0)) < ((4.75877285003662109)))*1.))))) \n    v[\"i17\"] = 0.100000*np.tanh(np.minimum((((-1.0*(((((data[\"pred_2\"]) > ((((6.0)) / 2.0)))*1.)))))), ((((data[\"t_mean\"]) - ((6.0))))))) \n    v[\"i18\"] = 0.100000*np.tanh(np.where(((data[\"t_sum_log\"]) - ((4.0)))>0, ((((5.0)) > (data[\"t_mean\"]))*1.), ((data[\"pred_8\"]) - ((5.0))) )) \n    v[\"i19\"] = 0.100000*np.tanh((-1.0*(((((((((((data[\"t_mean\"]) < ((4.0)))*1.)) * 2.0)) * 2.0)) * 2.0))))) \n    v[\"i20\"] = 0.100000*np.tanh(np.minimum((((((4.56055736541748047)) - (data[\"pred_1\"])))), ((np.minimum(((((data[\"t_sum_log\"]) - ((4.56055355072021484))))), (((((data[\"pred_4\"]) > (data[\"pred_1\"]))*1.)))))))) \n    v[\"i21\"] = 0.100000*np.tanh(np.minimum((((((((((data[\"t_mean\"]) * 2.0)) - ((9.0)))) < (3.0))*1.))), ((((((data[\"t_mean\"]) * 2.0)) - ((8.0))))))) \n    v[\"i22\"] = 0.100000*np.tanh(((np.minimum((((6.0))), ((np.where((6.0)>0, ((data[\"t_sum_log\"]) * (((data[\"t_sum_log\"]) / 2.0))), data[\"t_sum_log\"] ))))) - ((6.0)))) \n    v[\"i23\"] = 0.100000*np.tanh(np.minimum((((((data[\"t_sum_log\"]) > ((10.95762443542480469)))*1.))), ((((-3.0) - ((((10.95762443542480469)) * ((((3.0)) - (data[\"t_sum_log\"])))))))))) \n    v[\"i24\"] = 0.100000*np.tanh(((((np.minimum(((0.0)), ((((((data[\"t_sum_act\"]) + (((data[\"t_sum_act\"]) - ((6.0)))))) - (data[\"visitNumber\"])))))) * 2.0)) * 2.0)) \n    v[\"i25\"] = 0.100000*np.tanh(((np.minimum(((data[\"t_mean\"])), ((((np.minimum((((((data[\"t_mean\"]) < ((4.22032451629638672)))*1.))), ((((-3.0) + (data[\"t_sum_log\"])))))) * 2.0))))) * 2.0)) \n    v[\"i26\"] = 0.100000*np.tanh(((np.where(((-3.0) + (data[\"t_sum_log\"]))<0, ((((data[\"pred_2\"]) + (-2.0))) + (-2.0)), 0.0 )) * 2.0)) \n    v[\"i27\"] = 0.100000*np.tanh(np.minimum(((((((data[\"t_sum_log\"]) - ((4.0)))) - (((data[\"visitNumber\"]) / 2.0))))), (((((data[\"t_sum_act\"]) < (data[\"visitNumber\"]))*1.))))) \n    v[\"i28\"] = 0.100000*np.tanh(np.minimum(((((((((data[\"t_sum_act\"]) - (3.0))) * 2.0)) * 2.0))), ((np.minimum(((data[\"t_mean\"])), ((((data[\"t_sum_act\"]) - (data[\"t_mean\"]))))))))) \n    v[\"i29\"] = 0.100000*np.tanh(np.where(((-3.0) + (data[\"t_median\"]))<0, ((data[\"t_sum_log\"]) + (((((data[\"t_sum_log\"]) * 2.0)) - ((8.0))))), 0.0 )) \n    v[\"i30\"] = 0.100000*np.tanh(np.where(data[\"pred_1\"]>0, data[\"t_nb_sess\"], np.minimum(((((((data[\"pred_0\"]) - ((8.80866241455078125)))) * 2.0))), ((data[\"geoNetwork_metro\"]))) )) \n    v[\"i31\"] = 0.100000*np.tanh(np.where((((2.0) < (data[\"t_sum_log\"]))*1.)>0, (((((5.0)) - (data[\"totals_hits\"]))) - (data[\"visitNumber\"])), data[\"pred_5\"] )) \n    v[\"i32\"] = 0.100000*np.tanh(np.where((((2.0) < (((data[\"t_sum_act\"]) / 2.0)))*1.)>0, 0.0, ((((-3.0) + (data[\"t_sum_act\"]))) * 2.0) )) \n    v[\"i33\"] = 0.100000*np.tanh(((np.where(np.where(((data[\"t_sum_act\"]) - (2.0))>0, data[\"geoNetwork_continent\"], data[\"pred_5\"] )>0, (6.0), data[\"pred_1\"] )) - ((6.0)))) \n    v[\"i34\"] = 0.100000*np.tanh(np.minimum((((((data[\"pred_4\"]) > (-3.0))*1.))), ((((((((((data[\"t_sum_act\"]) * 2.0)) - (3.0))) * 2.0)) - (data[\"t_sum_act\"])))))) \n    v[\"i35\"] = 0.100000*np.tanh(np.minimum(((((((((data[\"t_sum_act\"]) - (3.0))) * 2.0)) + (data[\"t_nb_sess\"])))), (((((data[\"pred_5\"]) > ((0.0)))*1.))))) \n    v[\"i36\"] = 0.100000*np.tanh(((np.minimum(((((((-3.0) + (((data[\"t_sum_act\"]) * 2.0)))) * ((((2.0) > (data[\"t_sum_act\"]))*1.))))), ((data[\"t_median\"])))) * 2.0)) \n    v[\"i37\"] = 0.100000*np.tanh((((((((((data[\"t_sum_log\"]) < (3.0))*1.)) * 2.0)) * ((((((data[\"t_sum_log\"]) * 2.0)) + (-3.0))/2.0)))) * 2.0)) \n    v[\"i38\"] = 0.100000*np.tanh(np.minimum(((data[\"totals_pageviews\"])), (((((((((((1.0) > (data[\"t_median\"]))*1.)) * (data[\"pred_2\"]))) * (data[\"pred_3\"]))) * (data[\"pred_5\"])))))) \n    v[\"i39\"] = 0.100000*np.tanh(((((((data[\"pred_5\"]) * ((((2.0) > (data[\"t_sum_act\"]))*1.)))) * ((11.94209194183349609)))) * ((((2.0) > (data[\"t_sum_log\"]))*1.)))) \n    v[\"i40\"] = 0.100000*np.tanh(np.minimum(((((((((data[\"pred_6\"]) * 2.0)) * 2.0)) * ((((((data[\"t_sum_act\"]) < (((1.0) * 2.0)))*1.)) * 2.0))))), ((data[\"t_median\"])))) \n    v[\"i41\"] = 0.100000*np.tanh(np.where(((((data[\"t_nb_sess\"]) + (((data[\"t_sum_log\"]) * 2.0)))) + (-3.0))<0, -3.0, (((-2.0) > (data[\"t_nb_sess\"]))*1.) )) \n    v[\"i42\"] = 0.100000*np.tanh(np.where(((data[\"t_sum_log\"]) - (1.0))<0, ((data[\"t_sum_log\"]) - ((12.49324989318847656))), ((data[\"geoNetwork_continent\"]) - (np.tanh((data[\"totals_hits\"])))) )) \n    v[\"i43\"] = 0.100000*np.tanh(((np.minimum(((0.0)), ((((((((((data[\"t_sum_act\"]) * 2.0)) - (2.0))) * 2.0)) * 2.0))))) * 2.0)) \n    v[\"i44\"] = 0.100000*np.tanh(np.where(((data[\"t_sum_act\"]) - (1.0))<0, ((-3.0) * 2.0), ((np.tanh(((-1.0*((data[\"trafficSource_referralPath\"])))))) / 2.0) )) \n    v[\"i45\"] = 0.100000*np.tanh(((-3.0) * (np.where(data[\"t_mean\"]>0, ((((((data[\"t_sum_log\"]) < ((9.68791198730468750)))*1.)) < (data[\"pred_5\"]))*1.), (9.68791198730468750) )))) \n    v[\"i46\"] = 0.099951*np.tanh(((((((((-1.0) + (np.minimum(((np.tanh((((data[\"totals_pageviews\"]) * 2.0))))), ((data[\"t_sum_act\"])))))) * 2.0)) * 2.0)) * 2.0)) \n    v[\"i47\"] = 0.100000*np.tanh((((((((-1.0*(((((((((data[\"t_sum_act\"]) * 2.0)) + (-1.0))) < (data[\"pred_0\"]))*1.))))) * 2.0)) * 2.0)) * 2.0)) \n    v[\"i48\"] = 0.099902*np.tanh(((data[\"pred_4\"]) * (((((((((data[\"t_sum_act\"]) - (data[\"visitNumber\"]))) > (data[\"pred_4\"]))*1.)) > (((data[\"t_mean\"]) - (data[\"visitNumber\"]))))*1.)))) \n    v[\"i49\"] = 0.099756*np.tanh(((((-3.0) * 2.0)) * ((((np.tanh((data[\"t_sum_act\"]))) < ((((np.minimum(((data[\"totals_pageviews\"])), ((data[\"totals_hits\"])))) < (data[\"totals_hits\"]))*1.)))*1.)))) \n    v[\"i50\"] = 0.099951*np.tanh(np.minimum(((np.where(data[\"t_mean\"]>0, (((-3.0) < (data[\"pred_1\"]))*1.), -3.0 ))), ((np.where(data[\"geoNetwork_subContinent\"]>0, data[\"geoNetwork_subContinent\"], data[\"pred_1\"] ))))) \n    v[\"i51\"] = 0.100000*np.tanh(np.where(data[\"t_mean\"]<0, -3.0, (((np.maximum((((3.0))), ((((data[\"t_sum_log\"]) - (data[\"totals_pageviews\"])))))) > (data[\"t_mean\"]))*1.) )) \n    v[\"i52\"] = 0.100000*np.tanh((((((-1.0*(((((data[\"t_sum_log\"]) < (((np.maximum(((((data[\"visitNumber\"]) * 2.0))), ((data[\"t_sum_act\"])))) / 2.0)))*1.))))) * 2.0)) * 2.0)) \n    v[\"i53\"] = 0.100000*np.tanh(((np.where(np.minimum(((data[\"geoNetwork_subContinent\"])), ((data[\"geoNetwork_subContinent\"])))<0, data[\"pred_3\"], (((-1.0*(((((data[\"geoNetwork_subContinent\"]) > (data[\"t_sum_log\"]))*1.))))) * 2.0) )) * 2.0)) \n    v[\"i54\"] = 0.099658*np.tanh((-1.0*((np.where(data[\"t_sum_log\"]<0, (7.95942497253417969), np.where(data[\"pred_3\"]<0, (((data[\"t_sum_log\"]) > ((7.95942497253417969)))*1.), data[\"pred_2\"] ) ))))) \n    v[\"i55\"] = 0.099951*np.tanh((-1.0*((np.where(data[\"t_mean\"]<0, (6.22654199600219727), ((((((data[\"geoNetwork_metro\"]) > (0.0))*1.)) > (((data[\"geoNetwork_continent\"]) + (data[\"geoNetwork_metro\"]))))*1.) ))))) \n    v[\"i56\"] = 0.099951*np.tanh(((((((np.where(data[\"t_mean\"]>0, ((data[\"visitNumber\"]) * ((((data[\"totals_hits\"]) > (data[\"t_mean\"]))*1.))), -3.0 )) * 2.0)) * 2.0)) * 2.0)) \n    v[\"i57\"] = 0.099951*np.tanh((((((-1.0*((np.where(data[\"t_mean\"]>0, ((((((data[\"geoNetwork_continent\"]) > (data[\"geoNetwork_subContinent\"]))*1.)) < (data[\"geoNetwork_continent\"]))*1.), (9.77862071990966797) ))))) * 2.0)) * 2.0)) \n    v[\"i58\"] = 0.099609*np.tanh((((((-1.0*(((((np.tanh((data[\"geoNetwork_subContinent\"]))) < ((((data[\"t_mean\"]) < (((data[\"t_mean\"]) * (data[\"geoNetwork_subContinent\"]))))*1.)))*1.))))) * 2.0)) * 2.0)) \n    v[\"i59\"] = 0.100000*np.tanh(((np.where(data[\"geoNetwork_country\"]>0, ((np.where(data[\"t_sum_act\"]>0, (((data[\"pred_3\"]) < (data[\"pred_5\"]))*1.), data[\"pred_5\"] )) * 2.0), data[\"pred_3\"] )) * 2.0)) \n    v[\"i60\"] = 0.099951*np.tanh(np.where((((data[\"pred_0\"]) < (((data[\"t_nb_sess\"]) / 2.0)))*1.)>0, data[\"pred_4\"], (((((data[\"pred_0\"]) < (data[\"pred_4\"]))*1.)) * (-3.0)) )) \n    v[\"i61\"] = 0.100000*np.tanh(np.where(data[\"t_sum_log\"]<0, -3.0, (((np.maximum(((data[\"pred_3\"])), ((3.0)))) < (((data[\"totals_hits\"]) + (np.tanh((-2.0))))))*1.) )) \n    v[\"i62\"] = 0.100000*np.tanh(((((np.where(data[\"t_sum_log\"]<0, data[\"pred_3\"], (-1.0*(((((((data[\"trafficSource_referralPath\"]) * (data[\"t_sum_log\"]))) > (data[\"visitNumber\"]))*1.)))) )) * 2.0)) * 2.0)) \n    v[\"i63\"] = 0.100000*np.tanh(np.where(data[\"t_sum_act\"]<0, ((data[\"totals_hits\"]) + (-3.0)), (((data[\"t_median\"]) < (((((data[\"totals_pageviews\"]) + (-2.0))) * 2.0)))*1.) )) \n    v[\"i64\"] = 0.095457*np.tanh(np.where(data[\"t_mean\"]<0, (-1.0*(((9.0)))), (((data[\"t_mean\"]) < (np.where(data[\"t_sum_act\"]>0, (((9.0)) / 2.0), data[\"t_mean\"] )))*1.) )) \n    v[\"i65\"] = 0.100000*np.tanh(np.minimum(((data[\"t_median\"])), (((-1.0*((np.where(data[\"t_sum_log\"]<0, 3.0, (((data[\"geoNetwork_metro\"]) > (((data[\"t_mean\"]) / 2.0)))*1.) )))))))) \n    v[\"i66\"] = 0.100000*np.tanh(np.where(((data[\"geoNetwork_continent\"]) - (data[\"geoNetwork_subContinent\"]))<0, -3.0, np.where(data[\"t_sum_act\"]<0, -3.0, (((data[\"totals_pageviews\"]) > (data[\"t_mean\"]))*1.) ) )) \n    v[\"i67\"] = 0.100000*np.tanh(((-3.0) * (((((((np.tanh((((((data[\"totals_pageviews\"]) - (data[\"totals_hits\"]))) * 2.0)))) > (data[\"geoNetwork_country\"]))*1.)) > (data[\"geoNetwork_country\"]))*1.)))) \n    v[\"i68\"] = 0.099951*np.tanh(np.where((((data[\"pred_0\"]) + (data[\"t_sum_act\"]))/2.0)<0, -3.0, ((np.tanh(((((((data[\"pred_0\"]) < (data[\"t_median\"]))*1.)) / 2.0)))) / 2.0) )) \n    v[\"i69\"] = 0.100000*np.tanh((((((data[\"t_sum_log\"]) < (((np.maximum(((data[\"t_nb_sess\"])), ((data[\"totals_hits\"])))) * 2.0)))*1.)) * ((-1.0*(((((data[\"totals_pageviews\"]) < (data[\"totals_hits\"]))*1.))))))) \n    v[\"i70\"] = 0.100000*np.tanh(((((np.where(data[\"geoNetwork_continent\"]<0, data[\"visitNumber\"], (((np.tanh((data[\"t_sum_act\"]))) > ((((data[\"totals_hits\"]) + (data[\"visitNumber\"]))/2.0)))*1.) )) * 2.0)) * 2.0)) \n    v[\"i71\"] = 0.099902*np.tanh(((-3.0) * ((((((data[\"geoNetwork_metro\"]) / 2.0)) > (np.where(data[\"t_mean\"]>0, ((np.tanh((data[\"totals_hits\"]))) * 2.0), data[\"t_mean\"] )))*1.)))) \n    v[\"i72\"] = 0.100000*np.tanh(((((data[\"t_mean\"]) * ((((data[\"t_nb_sess\"]) < (((np.maximum(((((data[\"t_sum_log\"]) / 2.0))), ((data[\"pred_2\"])))) - (data[\"t_sum_log\"]))))*1.)))) * 2.0)) \n    v[\"i73\"] = 0.100000*np.tanh(np.where(data[\"pred_0\"]<0, (((-1.0*(((((data[\"pred_3\"]) > ((5.0)))*1.))))) * 2.0), (-1.0*(((((data[\"pred_0\"]) < (data[\"visitNumber\"]))*1.)))) )) \n    v[\"i74\"] = 0.083586*np.tanh(np.minimum(((data[\"t_sum_act\"])), (((((data[\"geoNetwork_metro\"]) > (np.maximum(((((data[\"pred_1\"]) / 2.0))), ((np.maximum((((1.40993750095367432))), ((data[\"visitNumber\"]))))))))*1.)))))\n    return Output(v.sum(axis=1))\n\ndef GP5(data):\n    v = pd.DataFrame()\n    v[\"i0\"] = 0.100000*np.tanh(((((((data[\"t_sum_act\"]) + (((data[\"t_sum_act\"]) - ((9.0)))))) + (((((data[\"t_sum_log\"]) - ((10.0)))) * 2.0)))) * 2.0)) \n    v[\"i1\"] = 0.100000*np.tanh(((data[\"t_sum_act\"]) - ((((6.94758796691894531)) - (((data[\"t_sum_log\"]) + (((((-2.0) * 2.0)) * 2.0)))))))) \n    v[\"i2\"] = 0.100000*np.tanh(((((((((data[\"t_sum_act\"]) + (((3.0) - ((10.0)))))) * 2.0)) * 2.0)) * 2.0)) \n    v[\"i3\"] = 0.100000*np.tanh(((((np.tanh((np.tanh((data[\"pred_3\"]))))) + (((((data[\"t_sum_act\"]) * 2.0)) - ((13.52748107910156250)))))) * 2.0)) \n    v[\"i4\"] = 0.100000*np.tanh(((((((data[\"t_sum_log\"]) - (np.maximum((((10.0))), ((data[\"t_median\"])))))) + (((data[\"t_sum_log\"]) / 2.0)))) + (-3.0))) \n    v[\"i5\"] = 0.100000*np.tanh(((((((((((data[\"t_mean\"]) - ((10.0)))) * 2.0)) + (data[\"t_sum_log\"]))) * 2.0)) * 2.0)) \n    v[\"i6\"] = 0.100000*np.tanh(((np.where(np.where(data[\"pred_0\"]>0, data[\"pred_2\"], ((data[\"sess_date_dow\"]) - (data[\"pred_0\"])) )>0, data[\"pred_0\"], data[\"t_sum_log\"] )) - ((9.53606986999511719)))) \n    v[\"i7\"] = 0.100000*np.tanh(np.minimum(((np.tanh((((data[\"t_median\"]) * 2.0))))), ((((((3.0) * 2.0)) * (((data[\"t_sum_act\"]) - ((6.01647520065307617))))))))) \n    v[\"i8\"] = 0.100000*np.tanh(((((((((((np.minimum(((data[\"t_sum_log\"])), (((11.64070796966552734))))) * 2.0)) - ((11.64070796966552734)))) * 2.0)) - (-1.0))) - (data[\"visitNumber\"]))) \n    v[\"i9\"] = 0.100000*np.tanh(((((np.where(data[\"t_median\"]>0, data[\"t_sum_log\"], ((((((((data[\"pred_2\"]) / 2.0)) * 2.0)) * 2.0)) * 2.0) )) - ((10.57200813293457031)))) * 2.0)) \n    v[\"i10\"] = 0.100000*np.tanh((((((7.02941036224365234)) - (data[\"t_sum_act\"]))) * (((data[\"t_sum_act\"]) - ((((((((7.02941036224365234)) - (data[\"t_sum_act\"]))) * 2.0)) * 2.0)))))) \n    v[\"i11\"] = 0.100000*np.tanh(np.minimum(((data[\"pred_1\"])), (((((((6.0)) / 2.0)) * (((data[\"t_sum_log\"]) - ((6.0))))))))) \n    v[\"i12\"] = 0.100000*np.tanh(((((data[\"t_sum_act\"]) - (np.maximum(((data[\"t_sum_act\"])), ((((((-3.0) * (np.tanh(((1.58075487613677979)))))) * (-2.0)))))))) * 2.0)) \n    v[\"i13\"] = 0.100000*np.tanh(((((data[\"t_sum_act\"]) * 2.0)) - (np.maximum(((((data[\"visitNumber\"]) * 2.0))), ((np.maximum(((((data[\"t_sum_act\"]) * 2.0))), (((10.78281593322753906)))))))))) \n    v[\"i14\"] = 0.100000*np.tanh(np.minimum((((((data[\"t_sum_act\"]) < ((6.0)))*1.))), ((((((np.minimum(((data[\"sess_date_hours\"])), ((((data[\"t_sum_act\"]) - ((5.0))))))) * 2.0)) * 2.0))))) \n    v[\"i15\"] = 0.100000*np.tanh(np.minimum((((((data[\"pred_1\"]) < ((((10.0)) - (data[\"t_sum_log\"]))))*1.))), ((((data[\"t_sum_log\"]) - ((((10.0)) - (data[\"t_sum_log\"])))))))) \n    v[\"i16\"] = 0.100000*np.tanh(np.minimum(((0.0)), ((((data[\"t_sum_log\"]) + (np.minimum(((((-1.0) - (data[\"visitNumber\"])))), ((((data[\"t_mean\"]) - ((9.99547958374023438)))))))))))) \n    v[\"i17\"] = 0.100000*np.tanh(((((-3.0) * 2.0)) + (np.minimum(((np.maximum(((data[\"pred_0\"])), ((data[\"t_sum_log\"]))))), ((((-3.0) * (-2.0)))))))) \n    v[\"i18\"] = 0.100000*np.tanh(((((((np.minimum(((((((5.0)) > (data[\"t_sum_act\"]))*1.))), ((((data[\"t_sum_act\"]) - ((4.49360561370849609))))))) * 2.0)) * 2.0)) * 2.0)) \n    v[\"i19\"] = 0.100000*np.tanh(((np.minimum(((((((((((data[\"t_mean\"]) / 2.0)) + (data[\"t_sum_log\"]))/2.0)) + (((data[\"t_sum_log\"]) * 2.0)))/2.0))), (((6.0))))) - ((6.0)))) \n    v[\"i20\"] = 0.100000*np.tanh(np.minimum(((0.0)), ((((np.where((((data[\"geoNetwork_networkDomain\"]) + ((1.0)))/2.0)<0, 0.0, ((data[\"t_sum_act\"]) * 2.0) )) - ((9.07516670227050781))))))) \n    v[\"i21\"] = 0.100000*np.tanh(np.minimum(((((((data[\"t_mean\"]) - ((3.69601702690124512)))) * ((7.17828702926635742))))), (((((3.0) > (((data[\"t_sum_act\"]) - (3.0))))*1.))))) \n    v[\"i22\"] = 0.100000*np.tanh(((np.minimum(((0.0)), ((((data[\"t_sum_log\"]) + (((((data[\"t_sum_log\"]) + (((-2.0) + (-3.0))))) * 2.0))))))) * 2.0)) \n    v[\"i23\"] = 0.100000*np.tanh(np.minimum(((0.0)), ((np.minimum(((((data[\"geoNetwork_networkDomain\"]) / 2.0))), (((((7.0)) * (((data[\"t_sum_act\"]) - ((((7.0)) / 2.0)))))))))))) \n    v[\"i24\"] = 0.100000*np.tanh(np.minimum(((0.0)), ((((((data[\"t_sum_log\"]) - ((7.0)))) + ((((data[\"t_sum_log\"]) + (data[\"t_sum_act\"]))/2.0))))))) \n    v[\"i25\"] = 0.100000*np.tanh(((np.minimum(((data[\"t_median\"])), ((((((((np.minimum(((0.0)), ((((data[\"t_sum_log\"]) + (-3.0)))))) * 2.0)) * 2.0)) * 2.0))))) * 2.0)) \n    v[\"i26\"] = 0.100000*np.tanh(((np.minimum(((((((((((data[\"t_sum_act\"]) + (-3.0))) * 2.0)) * 2.0)) * 2.0))), ((0.0)))) * ((9.0)))) \n    v[\"i27\"] = 0.100000*np.tanh(((((data[\"pred_2\"]) - ((5.99040746688842773)))) * ((((((np.maximum(((3.0)), ((((data[\"pred_2\"]) * 2.0))))) > (data[\"t_sum_log\"]))*1.)) * 2.0)))) \n    v[\"i28\"] = 0.100000*np.tanh(np.minimum(((((data[\"sess_date_dow\"]) / (-2.0)))), ((((((data[\"t_sum_log\"]) * (data[\"t_sum_log\"]))) - ((7.08554697036743164))))))) \n    v[\"i29\"] = 0.100000*np.tanh(np.minimum(((((((((data[\"t_sum_log\"]) * 2.0)) * 2.0)) - ((9.0))))), (((((((0.0) - (data[\"pred_0\"]))) > (data[\"geoNetwork_networkDomain\"]))*1.))))) \n    v[\"i30\"] = 0.100000*np.tanh(((((((data[\"t_sum_log\"]) + (-3.0))) * ((((2.0) > (((data[\"t_sum_act\"]) / 2.0)))*1.)))) * 2.0)) \n    v[\"i31\"] = 0.100000*np.tanh(np.minimum(((0.0)), ((((data[\"totals_pageviews\"]) - (np.where((((data[\"t_sum_act\"]) + (((-1.0) * 2.0)))/2.0)<0, (9.12072467803955078), data[\"totals_hits\"] ))))))) \n    v[\"i32\"] = 0.100000*np.tanh(np.minimum(((((((((((((((-2.0) + (data[\"t_sum_log\"]))) * 2.0)) * 2.0)) * 2.0)) * 2.0)) * 2.0))), ((0.0)))) \n    v[\"i33\"] = 0.100000*np.tanh(np.where((((-2.0) + (data[\"t_sum_log\"]))/2.0)<0, ((data[\"t_sum_log\"]) - ((10.0))), ((-1.0) / ((9.0))) )) \n    v[\"i34\"] = 0.100000*np.tanh(((np.minimum(((data[\"t_median\"])), ((((np.minimum((((((data[\"pred_2\"]) > (data[\"t_median\"]))*1.))), ((((data[\"t_sum_act\"]) + (-2.0)))))) * 2.0))))) * 2.0)) \n    v[\"i35\"] = 0.100000*np.tanh(((((np.where((((-3.0) + (data[\"t_sum_log\"]))/2.0)<0, ((((data[\"t_sum_log\"]) * 2.0)) + (-3.0)), 0.0 )) * 2.0)) * 2.0)) \n    v[\"i36\"] = 0.100000*np.tanh(((np.where(((-3.0) + (data[\"t_sum_act\"]))<0, -3.0, 0.0 )) * 2.0)) \n    v[\"i37\"] = 0.100000*np.tanh(np.minimum((((((data[\"t_mean\"]) < (3.0))*1.))), ((((((((((-3.0) + (((data[\"t_sum_act\"]) * 2.0)))) * 2.0)) * 2.0)) * 2.0))))) \n    v[\"i38\"] = 0.100000*np.tanh(((((np.minimum(((data[\"pred_1\"])), ((np.where(data[\"pred_2\"]>0, data[\"t_mean\"], -3.0 ))))) * 2.0)) * ((((data[\"t_mean\"]) < (2.0))*1.)))) \n    v[\"i39\"] = 0.100000*np.tanh(np.minimum((((0.0))), ((((data[\"t_median\"]) - ((((5.18295526504516602)) - (((data[\"t_sum_log\"]) * (((data[\"t_sum_act\"]) + (data[\"t_sum_log\"])))))))))))) \n    v[\"i40\"] = 0.100000*np.tanh(np.minimum(((((data[\"totals_pageviews\"]) - (((((2.0) / (data[\"t_mean\"]))) / (data[\"t_mean\"])))))), ((0.0)))) \n    v[\"i41\"] = 0.100000*np.tanh(((((((-3.0) * ((((1.0) > (data[\"t_sum_act\"]))*1.)))) * 2.0)) * 2.0)) \n    v[\"i42\"] = 0.100000*np.tanh(((-3.0) * ((((6.85303115844726562)) * ((((data[\"t_sum_log\"]) < (np.maximum(((data[\"visitNumber\"])), ((np.tanh((2.0)))))))*1.)))))) \n    v[\"i43\"] = 0.100000*np.tanh(np.minimum(((((data[\"t_sum_log\"]) - (2.0)))), ((np.minimum(((((2.0) / (data[\"t_sum_log\"])))), ((((data[\"t_sum_act\"]) - (data[\"t_mean\"]))))))))) \n    v[\"i44\"] = 0.100000*np.tanh(np.minimum(((((((2.0) - (data[\"visitNumber\"]))) / (data[\"t_sum_log\"])))), ((((((data[\"t_sum_log\"]) - (data[\"visitNumber\"]))) - (2.0)))))) \n    v[\"i45\"] = 0.099951*np.tanh(np.minimum(((np.minimum(((data[\"t_sum_act\"])), ((((1.0) / (data[\"t_sum_act\"]))))))), ((((data[\"sess_date_dom\"]) + (np.tanh((np.tanh((data[\"t_sum_log\"])))))))))) \n    v[\"i46\"] = 0.100000*np.tanh(np.minimum(((np.minimum(((data[\"t_mean\"])), (((((((3.0) / ((5.0)))) > (data[\"t_median\"]))*1.)))))), (((((3.95915246009826660)) / (data[\"t_mean\"])))))) \n    v[\"i47\"] = 0.100000*np.tanh(np.minimum(((np.minimum(((((data[\"visitNumber\"]) / (data[\"t_sum_act\"])))), ((((data[\"pred_1\"]) / (((data[\"t_sum_act\"]) / (data[\"visitNumber\"]))))))))), ((0.0)))) \n    v[\"i48\"] = 0.100000*np.tanh(np.where(data[\"t_mean\"]>0, ((-1.0) * (np.minimum((((((data[\"totals_hits\"]) > (data[\"totals_pageviews\"]))*1.))), ((np.tanh((data[\"totals_hits\"]))))))), -3.0 )) \n    v[\"i49\"] = 0.100000*np.tanh(np.minimum(((((3.0) / (((data[\"t_mean\"]) * 2.0))))), ((((data[\"t_sum_act\"]) + (((((data[\"t_sum_act\"]) * 2.0)) - (3.0)))))))) \n    v[\"i50\"] = 0.100000*np.tanh(np.where(((((data[\"t_nb_sess\"]) + (data[\"t_sum_log\"]))) + (data[\"t_sum_log\"]))>0, (((-3.0) > (data[\"t_nb_sess\"]))*1.), -3.0 )) \n    v[\"i51\"] = 0.100000*np.tanh(((np.minimum((((((((data[\"pred_0\"]) > (((data[\"t_sum_log\"]) * 2.0)))*1.)) / (data[\"trafficSource_source\"])))), ((((-1.0) + (data[\"t_sum_act\"])))))) * 2.0)) \n    v[\"i52\"] = 0.099902*np.tanh((((((((((data[\"totals_hits\"]) * (data[\"t_sum_act\"]))) < (1.0))*1.)) * (-3.0))) * 2.0)) \n    v[\"i53\"] = 0.100000*np.tanh(np.minimum((((((((2.0) < (data[\"totals_pageviews\"]))*1.)) + (data[\"t_median\"])))), (((((((data[\"t_median\"]) < (2.0))*1.)) / (data[\"t_mean\"])))))) \n    v[\"i54\"] = 0.099805*np.tanh((((((((data[\"t_sum_act\"]) * 2.0)) < (data[\"pred_0\"]))*1.)) * (((((-2.0) * 2.0)) * 2.0)))) \n    v[\"i55\"] = 0.099902*np.tanh(np.minimum((((((data[\"pred_1\"]) > (data[\"geoNetwork_city\"]))*1.))), ((((((((5.19124889373779297)) > (data[\"t_mean\"]))*1.)) / (np.tanh((data[\"t_mean\"])))))))) \n    v[\"i56\"] = 0.100000*np.tanh(np.where((((-1.0) + (((data[\"t_sum_log\"]) * 2.0)))/2.0)>0, ((((data[\"geoNetwork_networkDomain\"]) * (data[\"geoNetwork_networkDomain\"]))) / (-2.0)), -3.0 )) \n    v[\"i57\"] = 0.100000*np.tanh(((np.where(data[\"t_mean\"]>0, (((data[\"trafficSource_source\"]) > (np.tanh((np.tanh((np.tanh((data[\"totals_pageviews\"]))))))))*1.), (7.56566238403320312) )) / (-2.0))) \n    v[\"i58\"] = 0.099951*np.tanh(np.minimum(((((((3.0) * 2.0)) / (data[\"t_mean\"])))), ((np.minimum(((data[\"t_median\"])), (((((-1.0) > (data[\"sess_date_dom\"]))*1.)))))))) \n    v[\"i59\"] = 0.099951*np.tanh(np.minimum(((np.where(data[\"t_sum_log\"]<0, -3.0, ((data[\"channelGrouping\"]) * (np.tanh((data[\"t_nb_sess\"])))) ))), ((((data[\"channelGrouping\"]) * (data[\"channelGrouping\"])))))) \n    v[\"i60\"] = 0.093258*np.tanh(np.where(data[\"t_mean\"]>0, np.minimum(((np.tanh((np.tanh((data[\"pred_0\"])))))), ((((data[\"geoNetwork_networkDomain\"]) / (data[\"visitNumber\"]))))), -3.0 )) \n    v[\"i61\"] = 0.099951*np.tanh(np.minimum(((((((((data[\"t_sum_log\"]) * 2.0)) * 2.0)) - ((4.0))))), ((((((((data[\"visitNumber\"]) + (data[\"pred_0\"]))/2.0)) > ((9.99041843414306641)))*1.))))) \n    v[\"i62\"] = 0.097460*np.tanh((((((data[\"t_mean\"]) < (((-1.0) / (data[\"pred_1\"]))))*1.)) / (((np.tanh((data[\"t_mean\"]))) / 2.0)))) \n    v[\"i63\"] = 0.099951*np.tanh(np.where(((data[\"geoNetwork_networkDomain\"]) + (np.tanh((np.tanh((data[\"t_sum_log\"]))))))>0, ((-1.0) + (np.tanh((data[\"totals_pageviews\"])))), -3.0 )) \n    v[\"i64\"] = 0.093747*np.tanh(((((((data[\"t_sum_act\"]) + (((-3.0) * 2.0)))) * (((((((data[\"t_sum_log\"]) > ((10.0)))*1.)) > (data[\"sess_date_hours\"]))*1.)))) * 2.0)) \n    v[\"i65\"] = 0.100000*np.tanh(np.minimum((((((data[\"sess_date_hours\"]) < (((-2.0) * (data[\"visitNumber\"]))))*1.))), ((((((data[\"t_sum_log\"]) - (data[\"visitNumber\"]))) / (data[\"geoNetwork_country\"])))))) \n    v[\"i66\"] = 0.100000*np.tanh((((((data[\"geoNetwork_continent\"]) < ((((((((((data[\"trafficSource_referralPath\"]) + (data[\"geoNetwork_subContinent\"]))/2.0)) + (data[\"geoNetwork_subContinent\"]))/2.0)) + (1.0))/2.0)))*1.)) * (-3.0))) \n    v[\"i67\"] = 0.100000*np.tanh(np.where(data[\"t_mean\"]>0, np.where(data[\"geoNetwork_country\"]>0, (((((-2.0) / (data[\"geoNetwork_city\"]))) > (data[\"totals_pageviews\"]))*1.), -3.0 ), -3.0 )) \n    v[\"i68\"] = 0.099853*np.tanh(np.minimum(((np.where(data[\"geoNetwork_country\"]>0, np.where(data[\"t_median\"]>0, (((data[\"t_sum_log\"]) < (data[\"totals_pageviews\"]))*1.), data[\"sess_date_hours\"] ), -3.0 ))), ((data[\"geoNetwork_subContinent\"])))) \n    v[\"i69\"] = 0.100000*np.tanh(np.minimum(((0.0)), ((np.minimum(((data[\"totals_pageviews\"])), (((((((((data[\"totals_pageviews\"]) > (data[\"totals_hits\"]))*1.)) + (data[\"trafficSource_source\"]))) * 2.0)))))))) \n    v[\"i70\"] = 0.100000*np.tanh(((np.minimum(((((data[\"geoNetwork_continent\"]) - (data[\"geoNetwork_country\"])))), ((((data[\"geoNetwork_country\"]) - (np.where(data[\"t_median\"]>0, data[\"geoNetwork_country\"], data[\"trafficSource_source\"] ))))))) * 2.0)) \n    v[\"i71\"] = 0.100000*np.tanh(np.where(data[\"t_sum_log\"]>0, np.where(data[\"geoNetwork_subContinent\"]>0, (((data[\"t_sum_log\"]) > ((11.69831085205078125)))*1.), ((((data[\"trafficSource_source\"]) * 2.0)) * 2.0) ), -3.0 )) \n    v[\"i72\"] = 0.091060*np.tanh(np.minimum((((((((data[\"sess_date_hours\"]) < (((data[\"geoNetwork_networkDomain\"]) - ((((-3.0) + (data[\"t_mean\"]))/2.0)))))*1.)) / (data[\"geoNetwork_networkDomain\"])))), ((data[\"totals_hits\"])))) \n    v[\"i73\"] = 0.099951*np.tanh(np.minimum(((((2.0) / (data[\"geoNetwork_subContinent\"])))), (((((((((data[\"geoNetwork_country\"]) > (data[\"geoNetwork_subContinent\"]))*1.)) * (data[\"t_median\"]))) * (data[\"geoNetwork_city\"])))))) \n    v[\"i74\"] = 0.099951*np.tanh(np.minimum((((((((data[\"t_sum_log\"]) - (data[\"totals_pageviews\"]))) > (data[\"t_mean\"]))*1.))), ((((data[\"t_sum_log\"]) * ((10.0))))))) \n    v[\"i75\"] = 0.097557*np.tanh(np.minimum(((((data[\"t_sum_log\"]) * 2.0))), (((((((data[\"t_median\"]) + (((data[\"trafficSource_source\"]) * 2.0)))) < ((((data[\"t_median\"]) > (data[\"totals_pageviews\"]))*1.)))*1.))))) \n    v[\"i76\"] = 0.100000*np.tanh((((((((((((((4.0)) < (data[\"t_mean\"]))*1.)) / 2.0)) / (data[\"trafficSource_source\"]))) + (-1.0))/2.0)) / 2.0)) \n    v[\"i77\"] = 0.099902*np.tanh(((np.minimum(((data[\"t_median\"])), ((((np.minimum(((data[\"totals_pageviews\"])), (((((data[\"totals_pageviews\"]) > ((((8.65321922302246094)) / 2.0)))*1.))))) * ((8.65321922302246094))))))) * 2.0)) \n    v[\"i78\"] = 0.099218*np.tanh(((np.where(data[\"t_median\"]>0, data[\"channelGrouping\"], np.where(data[\"channelGrouping\"]>0, data[\"t_median\"], data[\"pred_0\"] ) )) * (np.minimum(((data[\"geoNetwork_country\"])), ((data[\"geoNetwork_region\"])))))) \n    v[\"i79\"] = 0.099853*np.tanh((((np.where(data[\"t_median\"]<0, 2.0, data[\"t_mean\"] )) < (np.minimum(((data[\"t_sum_log\"])), ((np.minimum(((data[\"t_sum_log\"])), ((data[\"totals_hits\"]))))))))*1.)) \n    v[\"i80\"] = 0.099951*np.tanh(((np.where(data[\"t_sum_act\"]>0, (((2.0) < (((((-2.0) / (data[\"trafficSource_source\"]))) / 2.0)))*1.), 2.0 )) * (-3.0))) \n    v[\"i81\"] = 0.058964*np.tanh((((data[\"geoNetwork_region\"]) > ((((data[\"geoNetwork_networkDomain\"]) < ((((((data[\"geoNetwork_country\"]) * ((11.84922599792480469)))) < (data[\"t_sum_log\"]))*1.)))*1.)))*1.)) \n    v[\"i82\"] = 0.097655*np.tanh(np.minimum(((((data[\"geoNetwork_country\"]) / (((-3.0) + (data[\"geoNetwork_networkDomain\"])))))), ((((((-1.0) / 2.0)) / (data[\"pred_0\"])))))) \n    v[\"i83\"] = 0.080459*np.tanh(np.minimum((((((data[\"geoNetwork_city\"]) < (data[\"t_median\"]))*1.))), ((np.minimum(((data[\"t_median\"])), ((np.minimum(((data[\"t_sum_act\"])), (((((data[\"t_sum_act\"]) > (data[\"t_mean\"]))*1.))))))))))) \n    v[\"i84\"] = 0.099951*np.tanh((((np.where(data[\"geoNetwork_city\"]<0, data[\"sess_date_dom\"], np.where(data[\"sess_date_dom\"]<0, ((data[\"sess_date_dom\"]) / (data[\"totals_hits\"])), data[\"sess_date_hours\"] ) )) < (-1.0))*1.)) \n    v[\"i85\"] = 0.099316*np.tanh((((((data[\"totals_pageviews\"]) < (data[\"totals_hits\"]))*1.)) / (np.minimum(((np.where(data[\"channelGrouping\"]>0, data[\"totals_hits\"], -2.0 ))), ((-1.0)))))) \n    v[\"i86\"] = 0.100000*np.tanh(((((((((((((data[\"geoNetwork_networkDomain\"]) < (0.0))*1.)) + (data[\"geoNetwork_city\"]))) > ((2.0)))*1.)) / (data[\"geoNetwork_city\"]))) / (data[\"geoNetwork_city\"]))) \n    v[\"i87\"] = 0.067904*np.tanh(np.minimum(((((data[\"t_sum_act\"]) - (data[\"t_mean\"])))), (((((1.0)) - (data[\"geoNetwork_city\"])))))) \n    v[\"i88\"] = 0.099902*np.tanh((((((-1.0) + ((((((data[\"sess_date_hours\"]) > (data[\"geoNetwork_country\"]))*1.)) / (data[\"sess_date_hours\"]))))/2.0)) / 2.0)) \n    v[\"i89\"] = 0.099951*np.tanh((((((((np.minimum(((data[\"pred_0\"])), (((((data[\"pred_0\"]) > (data[\"t_median\"]))*1.))))) < (((data[\"t_median\"]) / (data[\"pred_0\"]))))*1.)) / 2.0)) / 2.0)) \n    v[\"i90\"] = 0.099951*np.tanh(((np.where(data[\"t_sum_act\"]>0, (((data[\"geoNetwork_region\"]) > (((np.minimum(((data[\"sess_date_dom\"])), ((data[\"sess_date_hours\"])))) + (data[\"t_sum_log\"]))))*1.), -3.0 )) * 2.0)) \n    v[\"i91\"] = 0.099902*np.tanh(((((np.minimum(((np.minimum(((data[\"totals_hits\"])), (((((data[\"t_mean\"]) < (data[\"totals_pageviews\"]))*1.)))))), ((((data[\"totals_pageviews\"]) - (data[\"totals_hits\"])))))) * 2.0)) * 2.0)) \n    v[\"i92\"] = 0.099951*np.tanh((((((data[\"t_median\"]) > (np.where(data[\"geoNetwork_networkDomain\"]<0, data[\"geoNetwork_networkDomain\"], data[\"t_sum_act\"] )))*1.)) / (((((-3.0) + (data[\"t_mean\"]))) * 2.0)))) \n    v[\"i93\"] = 0.099951*np.tanh((((((((((((((data[\"geoNetwork_networkDomain\"]) / (data[\"pred_0\"]))) + (data[\"sess_date_dom\"]))) + (-1.0))) > (3.0))*1.)) * 2.0)) * 2.0)) \n    v[\"i94\"] = 0.099853*np.tanh(((((((np.minimum(((data[\"sess_date_dow\"])), ((((1.0) / (data[\"trafficSource_source\"])))))) / 2.0)) / 2.0)) / 2.0)) \n    v[\"i95\"] = 0.082169*np.tanh(np.tanh((((((data[\"sess_date_dow\"]) - (data[\"geoNetwork_country\"]))) / (((((data[\"sess_date_hours\"]) * (data[\"sess_date_hours\"]))) - (data[\"geoNetwork_country\"]))))))) \n    v[\"i96\"] = 0.099951*np.tanh((((((data[\"totals_hits\"]) < (((3.0) / 2.0)))*1.)) * (-1.0))) \n    v[\"i97\"] = 0.100000*np.tanh(np.where(data[\"t_sum_act\"]<0, -1.0, ((((((np.minimum(((-1.0)), ((data[\"sess_date_dow\"])))) / 2.0)) + (data[\"geoNetwork_country\"]))) / (data[\"geoNetwork_networkDomain\"])) )) \n    v[\"i98\"] = 0.099951*np.tanh(np.where(data[\"t_sum_act\"]<0, -3.0, ((((((6.35631465911865234)) - ((8.0)))) > (np.where(data[\"sess_date_dow\"]<0, 0.0, data[\"sess_date_hours\"] )))*1.) )) \n    v[\"i99\"] = 0.099951*np.tanh(((np.minimum(((((((((((data[\"trafficSource_referralPath\"]) + (2.0))/2.0)) > (data[\"totals_hits\"]))*1.)) * (data[\"trafficSource_referralPath\"])))), ((data[\"totals_hits\"])))) * 2.0))\n\n    return Output(v.sum(axis=1))\n    \ndef GP6(data):\n    v = pd.DataFrame()\n    v[\"i0\"] = 0.100000*np.tanh(((((data[\"t_mean\"]) - ((9.82170104980468750)))) + (((data[\"t_mean\"]) - (((((data[\"t_sum_log\"]) - ((9.04127788543701172)))) * (-3.0))))))) \n    v[\"i1\"] = 0.100000*np.tanh((((((14.27411556243896484)) * (((data[\"t_mean\"]) + (((((data[\"t_sum_log\"]) + (((((-3.0) * 2.0)) * 2.0)))) * 2.0)))))) * 2.0)) \n    v[\"i2\"] = 0.100000*np.tanh(((((((((np.where(data[\"pred_1\"] < -99998, data[\"pred_1\"], ((data[\"t_mean\"]) - ((7.67527198791503906))) )) * 2.0)) * 2.0)) * 2.0)) + (3.0))) \n    v[\"i3\"] = 0.100000*np.tanh(((-3.0) + (((((((data[\"t_sum_log\"]) + (((data[\"t_mean\"]) - ((((7.0)) * 2.0)))))) * ((7.0)))) * 2.0)))) \n    v[\"i4\"] = 0.100000*np.tanh(((((data[\"t_sum_log\"]) - (np.maximum(((np.maximum((((6.96472883224487305))), ((((((data[\"visitNumber\"]) * 2.0)) - (data[\"t_sum_log\"]))))))), ((data[\"t_median\"])))))) * 2.0)) \n    v[\"i5\"] = 0.100000*np.tanh(((((((1.0) + (((((1.0) + (((((data[\"t_mean\"]) - ((7.40952205657958984)))) * 2.0)))) * 2.0)))) * 2.0)) * 2.0)) \n    v[\"i6\"] = 0.100000*np.tanh(((np.where(((data[\"t_sum_log\"]) - ((9.06132984161376953)))>0, ((data[\"t_sum_log\"]) - ((9.06132698059082031))), ((data[\"t_median\"]) - ((9.06132698059082031))) )) * 2.0)) \n    v[\"i7\"] = 0.100000*np.tanh(((np.minimum(((((((((data[\"t_sum_act\"]) - (((3.0) * 2.0)))) * 2.0)) * 2.0))), (((((data[\"t_sum_act\"]) > (data[\"t_mean\"]))*1.))))) * 2.0)) \n    v[\"i8\"] = 0.100000*np.tanh(np.minimum(((((((data[\"t_sum_act\"]) - ((((12.94684314727783203)) / 2.0)))) * 2.0))), (((((data[\"pred_3\"]) > (((((data[\"t_sum_act\"]) / 2.0)) / 2.0)))*1.))))) \n    v[\"i9\"] = 0.100000*np.tanh(((data[\"t_sum_log\"]) + (((((((((((data[\"t_sum_log\"]) + (((-3.0) * 2.0)))) * 2.0)) * 2.0)) - (data[\"visitNumber\"]))) * 2.0)))) \n    v[\"i10\"] = 0.100000*np.tanh(((((((((((((((-3.0) * 2.0)) + (data[\"sess_date_dom\"]))) + (data[\"t_sum_log\"]))) - ((5.83093786239624023)))) * 2.0)) * 2.0)) * 2.0)) \n    v[\"i11\"] = 0.100000*np.tanh(((((-3.0) + (((data[\"t_sum_log\"]) + (((((((data[\"t_sum_act\"]) + (((-3.0) * 2.0)))) * 2.0)) * 2.0)))))) * 2.0)) \n    v[\"i12\"] = 0.100000*np.tanh((((((10.66844654083251953)) * ((((((10.66844272613525391)) * (((data[\"t_sum_log\"]) - ((10.66844272613525391)))))) - (data[\"t_sum_log\"]))))) - ((10.66844272613525391)))) \n    v[\"i13\"] = 0.100000*np.tanh(((((((data[\"t_sum_log\"]) + (((data[\"t_sum_act\"]) - (np.maximum(((np.maximum((((10.0))), ((data[\"pred_0\"]))))), (((10.0))))))))) * 2.0)) * 2.0)) \n    v[\"i14\"] = 0.100000*np.tanh(((-3.0) + (np.minimum(((np.minimum(((((data[\"t_mean\"]) + (((data[\"t_sum_act\"]) - ((9.0))))))), (((2.37444567680358887)))))), (((2.37444567680358887))))))) \n    v[\"i15\"] = 0.100000*np.tanh((((((14.27473449707031250)) * (((np.where(data[\"sess_date_hours\"]>0, ((data[\"t_sum_act\"]) - ((3.87348985671997070))), -1.0 )) + (-1.0))))) * 2.0)) \n    v[\"i16\"] = 0.100000*np.tanh(np.minimum(((data[\"geoNetwork_networkDomain\"])), ((((np.maximum(((data[\"t_median\"])), ((data[\"pred_0\"])))) - (np.maximum((((((12.08827495574951172)) - (data[\"pred_0\"])))), ((data[\"pred_0\"]))))))))) \n    v[\"i17\"] = 0.100000*np.tanh(np.minimum(((((data[\"t_sum_log\"]) - ((((10.0)) / 2.0))))), ((((2.0) - (((data[\"t_sum_act\"]) - ((((10.0)) / 2.0))))))))) \n    v[\"i18\"] = 0.100000*np.tanh(((((((((np.minimum((((((data[\"t_sum_log\"]) < ((5.0)))*1.))), ((((data[\"t_sum_log\"]) - ((4.0))))))) * 2.0)) * 2.0)) * 2.0)) * 2.0)) \n    v[\"i19\"] = 0.100000*np.tanh(np.minimum((((((6.83367300033569336)) - (data[\"t_mean\"])))), ((np.minimum(((data[\"pred_1\"])), ((((data[\"t_mean\"]) - ((((6.93451642990112305)) - (data[\"t_mean\"]))))))))))) \n    v[\"i20\"] = 0.100000*np.tanh(np.minimum((((((-1.0) + (2.0))/2.0))), ((((((((((((data[\"t_mean\"]) - ((4.01557540893554688)))) * 2.0)) * 2.0)) * 2.0)) * 2.0))))) \n    v[\"i21\"] = 0.100000*np.tanh(np.minimum(((0.0)), (((((((data[\"t_sum_log\"]) * 2.0)) + (((data[\"t_sum_log\"]) - ((((12.81262588500976562)) + (data[\"visitNumber\"]))))))/2.0))))) \n    v[\"i22\"] = 0.100000*np.tanh(((np.minimum(((((data[\"t_sum_act\"]) - ((((2.0)) * 2.0))))), (((((((data[\"t_sum_act\"]) - (3.0))) < ((2.0)))*1.))))) * 2.0)) \n    v[\"i23\"] = 0.100000*np.tanh(np.minimum(((0.0)), ((((((((np.where(data[\"pred_0\"]>0, data[\"pred_0\"], data[\"t_mean\"] )) - (3.0))) * 2.0)) * 2.0))))) \n    v[\"i24\"] = 0.100000*np.tanh((((10.90211677551269531)) * ((((8.0)) * (((-3.0) + (np.minimum(((3.0)), ((np.minimum(((data[\"t_sum_act\"])), ((data[\"t_sum_log\"]))))))))))))) \n    v[\"i25\"] = 0.100000*np.tanh(((data[\"t_sum_log\"]) - (np.where(((-3.0) + ((((data[\"t_sum_log\"]) + (data[\"t_mean\"]))/2.0)))<0, (7.14634370803833008), data[\"t_sum_log\"] )))) \n    v[\"i26\"] = 0.100000*np.tanh(((((((np.where(((data[\"t_mean\"]) - (3.0))>0, 3.0, data[\"t_sum_log\"] )) - (3.0))) * 2.0)) * 2.0)) \n    v[\"i27\"] = 0.100000*np.tanh((((((((np.minimum(((((((data[\"t_mean\"]) * 2.0)) * 2.0))), ((data[\"t_sum_act\"])))) < (3.0))*1.)) * (-3.0))) * ((6.24691629409790039)))) \n    v[\"i28\"] = 0.100000*np.tanh((((((((((((data[\"t_sum_act\"]) / 2.0)) < (2.0))*1.)) * (((((data[\"t_sum_act\"]) * 2.0)) - ((5.52667522430419922)))))) * 2.0)) * 2.0)) \n    v[\"i29\"] = 0.100000*np.tanh(np.minimum(((((np.minimum(((0.0)), (((((((((data[\"t_sum_log\"]) - ((6.83591127395629883)))) + (data[\"t_sum_act\"]))/2.0)) * 2.0))))) * 2.0))), ((data[\"t_median\"])))) \n    v[\"i30\"] = 0.100000*np.tanh(((np.where(np.where(((-2.0) + (data[\"t_sum_act\"]))>0, data[\"t_mean\"], -3.0 )>0, 0.0, -3.0 )) * 2.0)) \n    v[\"i31\"] = 0.100000*np.tanh((((((data[\"t_sum_log\"]) < (2.0))*1.)) * (((((-3.0) * 2.0)) * 2.0)))) \n    v[\"i32\"] = 0.100000*np.tanh((((3.0) + (np.where((((data[\"t_sum_act\"]) + (-2.0))/2.0)<0, ((-2.0) * ((9.0))), -3.0 )))/2.0)) \n    v[\"i33\"] = 0.100000*np.tanh(np.minimum(((np.tanh((((((((data[\"trafficSource_source\"]) < (data[\"pred_2\"]))*1.)) > (data[\"sess_date_hours\"]))*1.))))), ((((((-2.0) + (data[\"t_sum_log\"]))) * 2.0))))) \n    v[\"i34\"] = 0.100000*np.tanh(((np.minimum((((((data[\"t_median\"]) > ((10.0)))*1.))), ((((np.minimum(((data[\"t_median\"])), ((((data[\"t_sum_act\"]) + (-2.0)))))) * 2.0))))) * 2.0)) \n    v[\"i35\"] = 0.100000*np.tanh(((np.minimum(((((np.minimum(((((data[\"t_sum_act\"]) - ((2.0))))), ((((((data[\"geoNetwork_country\"]) * 2.0)) * 2.0))))) * 2.0))), ((0.0)))) * 2.0)) \n    v[\"i36\"] = 0.100000*np.tanh(((((np.minimum((((((np.maximum(((data[\"t_median\"])), ((data[\"t_median\"])))) < (data[\"pred_2\"]))*1.))), ((((data[\"t_sum_log\"]) + (-2.0)))))) * 2.0)) * 2.0)) \n    v[\"i37\"] = 0.100000*np.tanh(np.minimum(((0.0)), (((((((-3.0) - (((data[\"pred_2\"]) * (((-3.0) + (data[\"t_median\"]))))))) + (data[\"t_sum_log\"]))/2.0))))) \n    v[\"i38\"] = 0.100000*np.tanh(((np.where(((-3.0) + (data[\"t_sum_log\"]))<0, ((((-3.0) + (data[\"t_sum_log\"]))) + (data[\"t_mean\"])), 0.0 )) * 2.0)) \n    v[\"i39\"] = 0.100000*np.tanh(np.minimum(((((-3.0) + (data[\"t_mean\"])))), ((np.where(data[\"trafficSource_source\"]<0, -1.0, ((3.0) - (data[\"visitNumber\"])) ))))) \n    v[\"i40\"] = 0.100000*np.tanh(((((-3.0) * 2.0)) * ((((np.tanh((((1.0) * 2.0)))) > (data[\"t_sum_log\"]))*1.)))) \n    v[\"i41\"] = 0.099951*np.tanh(np.where(((data[\"t_mean\"]) + (data[\"pred_2\"]))>0, (((data[\"totals_hits\"]) < (data[\"totals_pageviews\"]))*1.), ((data[\"totals_pageviews\"]) + (((-3.0) * 2.0))) )) \n    v[\"i42\"] = 0.100000*np.tanh(((((((((data[\"t_sum_log\"]) + (((data[\"t_sum_log\"]) + (-2.0))))) * ((((data[\"t_mean\"]) < (2.0))*1.)))) * 2.0)) * 2.0)) \n    v[\"i43\"] = 0.100000*np.tanh(np.minimum(((((2.0) - (np.where(data[\"t_mean\"]<0, (8.0), data[\"visitNumber\"] ))))), (((((data[\"totals_pageviews\"]) > (data[\"totals_hits\"]))*1.))))) \n    v[\"i44\"] = 0.100000*np.tanh(np.minimum(((((data[\"t_sum_act\"]) + (((-3.0) + (data[\"t_sum_act\"])))))), (((((((data[\"t_mean\"]) < (data[\"t_sum_act\"]))*1.)) - (data[\"trafficSource_source\"])))))) \n    v[\"i45\"] = 0.100000*np.tanh(np.where(data[\"t_mean\"]<0, -3.0, np.where(data[\"geoNetwork_networkDomain\"]>0, ((np.tanh((data[\"geoNetwork_networkDomain\"]))) - (data[\"geoNetwork_networkDomain\"])), (0.20732049643993378) ) )) \n    v[\"i46\"] = 0.100000*np.tanh((((((((((1.0)) + (data[\"t_sum_log\"]))/2.0)) > (data[\"t_sum_act\"]))*1.)) * (((-3.0) * 2.0)))) \n    v[\"i47\"] = 0.100000*np.tanh(((-3.0) * ((((np.tanh((((-1.0) + (data[\"t_sum_log\"]))))) < ((((data[\"geoNetwork_networkDomain\"]) < (np.tanh((-1.0))))*1.)))*1.)))) \n    v[\"i48\"] = 0.100000*np.tanh(np.minimum(((((((data[\"t_sum_act\"]) - (3.0))) + (data[\"t_sum_act\"])))), ((((((((data[\"pred_0\"]) + (data[\"sess_date_hours\"]))/2.0)) > (data[\"t_sum_log\"]))*1.))))) \n    v[\"i49\"] = 0.100000*np.tanh(np.where(((data[\"t_median\"]) + (data[\"pred_1\"])) < -99998, data[\"pred_1\"], np.where(data[\"t_mean\"]<0, -3.0, (((data[\"pred_1\"]) > (data[\"pred_1\"]))*1.) ) )) \n    v[\"i50\"] = 0.100000*np.tanh(((((-3.0) * ((((np.maximum(((data[\"visitNumber\"])), ((np.tanh((np.tanh((np.tanh((data[\"t_mean\"])))))))))) > (data[\"t_sum_log\"]))*1.)))) * 2.0)) \n    v[\"i51\"] = 0.099951*np.tanh(np.where(data[\"t_sum_log\"]<0, ((-3.0) * 2.0), ((((((data[\"t_sum_log\"]) < (data[\"pred_0\"]))*1.)) < ((((-3.0) + (data[\"visitNumber\"]))/2.0)))*1.) )) \n    v[\"i52\"] = 0.099902*np.tanh(np.where(np.where(data[\"t_mean\"]>0, (((data[\"geoNetwork_networkDomain\"]) + ((1.0)))/2.0), -3.0 )>0, (((data[\"t_mean\"]) < (data[\"totals_pageviews\"]))*1.), -3.0 )) \n    v[\"i53\"] = 0.100000*np.tanh(np.where(data[\"t_sum_log\"]<0, -3.0, np.minimum(((data[\"pred_0\"])), ((np.minimum(((data[\"t_median\"])), ((((((9.0)) < (((data[\"pred_0\"]) / 2.0)))*1.))))))) )) \n    v[\"i54\"] = 0.100000*np.tanh(np.where(((((data[\"t_nb_sess\"]) / 2.0)) + (data[\"t_sum_log\"]))<0, ((-2.0) * 2.0), ((data[\"sess_date_dow\"]) * (((data[\"t_nb_sess\"]) / 2.0))) )) \n    v[\"i55\"] = 0.100000*np.tanh(((-1.0) + (((np.tanh((((data[\"t_sum_act\"]) * 2.0)))) - ((((data[\"totals_hits\"]) > (data[\"totals_pageviews\"]))*1.)))))) \n    v[\"i56\"] = 0.099609*np.tanh(np.where(data[\"t_mean\"]>0, (((data[\"t_mean\"]) < (((((((3.0) + (data[\"t_sum_log\"]))/2.0)) + (data[\"t_sum_log\"]))/2.0)))*1.), -2.0 )) \n    v[\"i57\"] = 0.100000*np.tanh(((((((((np.tanh((np.tanh((((data[\"totals_hits\"]) / 2.0)))))) < (data[\"trafficSource_source\"]))*1.)) > (np.tanh((data[\"t_sum_act\"]))))*1.)) * (-3.0))) \n    v[\"i58\"] = 0.099951*np.tanh(np.minimum(((((((((((-1.0) + (data[\"t_sum_log\"]))) + (data[\"t_sum_log\"]))) * 2.0)) * 2.0))), (((((-2.0) > (data[\"sess_date_hours\"]))*1.))))) \n    v[\"i59\"] = 0.100000*np.tanh(((np.minimum(((0.0)), ((np.minimum(((((data[\"t_sum_log\"]) - (data[\"visitNumber\"])))), ((((((data[\"t_sum_log\"]) - (1.0))) * 2.0)))))))) * 2.0)) \n    v[\"i60\"] = 0.100000*np.tanh(np.minimum(((np.where(data[\"t_sum_log\"]<0, -3.0, ((((((((2.0)) > (data[\"totals_pageviews\"]))*1.)) / 2.0)) / 2.0) ))), ((data[\"totals_pageviews\"])))) \n    v[\"i61\"] = 0.099951*np.tanh(np.where(data[\"t_sum_act\"]<0, -2.0, ((np.minimum(((data[\"sess_date_dom\"])), (((((data[\"pred_0\"]) < (data[\"t_median\"]))*1.))))) / 2.0) )) \n    v[\"i62\"] = 0.099951*np.tanh(np.where(data[\"t_sum_act\"]>0, np.minimum((((((data[\"visitNumber\"]) < (np.tanh((data[\"sess_date_dom\"]))))*1.))), (((((data[\"visitNumber\"]) < (data[\"trafficSource_source\"]))*1.)))), -3.0 )) \n    v[\"i63\"] = 0.100000*np.tanh(np.minimum((((((((data[\"pred_1\"]) * (((data[\"totals_hits\"]) - (data[\"totals_pageviews\"]))))) + (data[\"totals_pageviews\"]))/2.0))), (((((data[\"sess_date_dom\"]) < (-1.0))*1.))))) \n    v[\"i64\"] = 0.099951*np.tanh(np.minimum(((np.where(data[\"t_sum_act\"]<0, -3.0, 0.0 ))), ((((data[\"totals_hits\"]) - (np.maximum(((data[\"totals_pageviews\"])), ((data[\"sess_date_hours\"]))))))))) \n    v[\"i65\"] = 0.100000*np.tanh(np.minimum(((((-1.0) + (data[\"t_sum_log\"])))), ((np.minimum(((data[\"totals_pageviews\"])), (((((2.0) > (((data[\"geoNetwork_city\"]) + (data[\"totals_pageviews\"]))))*1.)))))))) \n    v[\"i66\"] = 0.100000*np.tanh(((np.minimum(((data[\"totals_hits\"])), ((np.minimum(((((data[\"totals_pageviews\"]) - (data[\"totals_hits\"])))), ((((data[\"totals_hits\"]) - (data[\"totals_pageviews\"]))))))))) * 2.0)) \n    v[\"i67\"] = 0.099951*np.tanh(((((((((4.0)) < (data[\"totals_pageviews\"]))*1.)) * (((data[\"sess_date_hours\"]) + (((data[\"pred_0\"]) - (data[\"totals_pageviews\"]))))))) * (data[\"t_mean\"]))) \n    v[\"i68\"] = 0.099707*np.tanh(np.where(data[\"t_mean\"]<0, ((-3.0) * 2.0), ((np.minimum((((((1.0) < (data[\"geoNetwork_region\"]))*1.))), ((data[\"totals_hits\"])))) / 2.0) )) \n    v[\"i69\"] = 0.082218*np.tanh((((((np.where(data[\"sess_date_hours\"]>0, data[\"sess_date_hours\"], data[\"pred_1\"] )) > (2.0))*1.)) - ((((data[\"sess_date_hours\"]) < (((data[\"geoNetwork_city\"]) / 2.0)))*1.)))) \n    v[\"i70\"] = 0.078065*np.tanh(np.where(data[\"t_sum_log\"]>0, np.tanh((((np.where(data[\"sess_date_hours\"]>0, ((3.0) * 2.0), data[\"t_mean\"] )) - (data[\"t_sum_act\"])))), -3.0 )) \n    v[\"i71\"] = 0.100000*np.tanh(np.where(data[\"geoNetwork_country\"]>0, (((((1.0) < (((data[\"geoNetwork_city\"]) * ((((data[\"geoNetwork_city\"]) + (data[\"geoNetwork_region\"]))/2.0)))))*1.)) / 2.0), -3.0 )) \n    v[\"i72\"] = 0.100000*np.tanh(((np.where(data[\"geoNetwork_subContinent\"]<0, -3.0, np.where((((data[\"geoNetwork_subContinent\"]) > (data[\"geoNetwork_country\"]))*1.)>0, 0.0, -3.0 ) )) * 2.0)) \n    v[\"i73\"] = 0.099951*np.tanh((((11.99727344512939453)) * (np.minimum(((data[\"geoNetwork_subContinent\"])), ((np.minimum(((data[\"geoNetwork_country\"])), ((np.minimum(((data[\"t_sum_log\"])), (((((data[\"pred_1\"]) > (data[\"t_sum_log\"]))*1.))))))))))))) \n    v[\"i74\"] = 0.099951*np.tanh(np.minimum(((((data[\"geoNetwork_networkDomain\"]) * (np.tanh((((((data[\"pred_1\"]) + (data[\"sess_date_dom\"]))) + (((data[\"geoNetwork_networkDomain\"]) * 2.0))))))))), ((0.0)))) \n    v[\"i75\"] = 0.099707*np.tanh(((data[\"totals_pageviews\"]) * (np.maximum((((((data[\"totals_pageviews\"]) > (data[\"t_mean\"]))*1.))), (((((((data[\"t_sum_log\"]) - (data[\"totals_pageviews\"]))) > (data[\"t_mean\"]))*1.))))))) \n    v[\"i76\"] = 0.099902*np.tanh(np.minimum((((((data[\"geoNetwork_country\"]) < (data[\"sess_date_hours\"]))*1.))), ((((data[\"geoNetwork_country\"]) - ((((data[\"geoNetwork_country\"]) < ((((data[\"geoNetwork_subContinent\"]) < (data[\"sess_date_hours\"]))*1.)))*1.))))))) \n    v[\"i77\"] = 0.099902*np.tanh(np.where(data[\"sess_date_dom\"]>0, ((((0.05049348995089531)) > (data[\"sess_date_dom\"]))*1.), (((((data[\"sess_date_hours\"]) > (data[\"geoNetwork_country\"]))*1.)) / 2.0) )) \n    v[\"i78\"] = 0.100000*np.tanh(np.minimum(((np.where(data[\"geoNetwork_subContinent\"]<0, -1.0, data[\"t_median\"] ))), (((((data[\"sess_date_hours\"]) < (((data[\"geoNetwork_networkDomain\"]) - (2.0))))*1.))))) \n    v[\"i79\"] = 0.100000*np.tanh(((np.minimum(((((data[\"t_mean\"]) - (((((data[\"geoNetwork_region\"]) + (data[\"geoNetwork_city\"]))) + (((data[\"geoNetwork_region\"]) * 2.0))))))), ((0.0)))) * 2.0)) \n    v[\"i80\"] = 0.100000*np.tanh(np.minimum((((((((data[\"trafficSource_referralPath\"]) * 2.0)) > (-1.0))*1.))), ((((((-2.0) * (((data[\"trafficSource_referralPath\"]) * 2.0)))) + (data[\"geoNetwork_continent\"])))))) \n    v[\"i81\"] = 0.099951*np.tanh((((((((((data[\"trafficSource_source\"]) > (((data[\"totals_hits\"]) - (((data[\"trafficSource_referralPath\"]) / 2.0)))))*1.)) * (data[\"sess_date_hours\"]))) * (data[\"t_mean\"]))) * 2.0)) \n    v[\"i82\"] = 0.099853*np.tanh(((((((np.where(data[\"geoNetwork_subContinent\"]<0, data[\"geoNetwork_city\"], (((3.0) < ((((data[\"geoNetwork_region\"]) + (data[\"totals_pageviews\"]))/2.0)))*1.) )) * 2.0)) * 2.0)) * 2.0)) \n    v[\"i83\"] = 0.099951*np.tanh(((np.minimum(((data[\"totals_hits\"])), (((((data[\"t_median\"]) < ((((data[\"totals_hits\"]) + (((-2.0) - (np.tanh((data[\"geoNetwork_country\"]))))))/2.0)))*1.))))) * 2.0)) \n    v[\"i84\"] = 0.099902*np.tanh(((((((((((data[\"trafficSource_referralPath\"]) < (data[\"totals_pageviews\"]))*1.)) - (data[\"totals_pageviews\"]))) > (data[\"trafficSource_source\"]))*1.)) - ((((1.0) < (data[\"geoNetwork_country\"]))*1.)))) \n    v[\"i85\"] = 0.081974*np.tanh((((((data[\"totals_pageviews\"]) > (data[\"t_sum_log\"]))*1.)) - (np.where(data[\"trafficSource_referralPath\"]<0, (((((3.0) / 2.0)) > (data[\"totals_pageviews\"]))*1.), data[\"geoNetwork_city\"] )))) \n    v[\"i86\"] = 0.070982*np.tanh(((np.minimum(((((data[\"geoNetwork_city\"]) * (data[\"trafficSource_medium\"])))), ((((((((data[\"trafficSource_source\"]) < (((data[\"trafficSource_medium\"]) * 2.0)))*1.)) < (data[\"trafficSource_source\"]))*1.))))) * 2.0)) \n    v[\"i87\"] = 0.099951*np.tanh((((((np.maximum(((((data[\"totals_pageviews\"]) * 2.0))), ((data[\"t_mean\"])))) > (np.maximum(((((data[\"totals_pageviews\"]) + (data[\"totals_hits\"])))), ((data[\"pred_0\"])))))*1.)) / 2.0)) \n    v[\"i88\"] = 0.076160*np.tanh((((((data[\"pred_0\"]) < (np.minimum(((data[\"t_sum_log\"])), ((2.0)))))*1.)) * (((data[\"t_median\"]) * (np.minimum(((data[\"pred_0\"])), ((data[\"geoNetwork_region\"])))))))) \n    v[\"i89\"] = 0.099951*np.tanh(np.where(((data[\"channelGrouping\"]) + (data[\"t_sum_act\"]))>0, ((((np.minimum(((data[\"geoNetwork_city\"])), ((data[\"sess_date_dow\"])))) / 2.0)) * (data[\"channelGrouping\"])), -3.0 )) \n    v[\"i90\"] = 0.085344*np.tanh(((np.minimum(((data[\"trafficSource_source\"])), ((np.where(data[\"trafficSource_source\"]<0, -1.0, ((((((data[\"geoNetwork_city\"]) > (data[\"geoNetwork_region\"]))*1.)) > (data[\"trafficSource_source\"]))*1.) ))))) / 2.0)) \n    v[\"i91\"] = 0.099951*np.tanh(((((((((((data[\"sess_date_dom\"]) > (data[\"geoNetwork_country\"]))*1.)) > ((((-1.0) + (data[\"t_mean\"]))/2.0)))*1.)) * 2.0)) * (data[\"t_sum_log\"]))) \n    v[\"i92\"] = 0.099853*np.tanh(np.where(data[\"t_mean\"]<0, -3.0, np.minimum(((data[\"totals_hits\"])), (((((data[\"trafficSource_referralPath\"]) > (data[\"totals_hits\"]))*1.)))) )) \n    v[\"i93\"] = 0.099511*np.tanh(((((((data[\"sess_date_hours\"]) < (data[\"sess_date_dow\"]))*1.)) < (np.tanh((np.where(data[\"sess_date_dom\"]<0, (((data[\"sess_date_dow\"]) < (-1.0))*1.), data[\"sess_date_dow\"] )))))*1.)) \n    v[\"i94\"] = 0.099853*np.tanh(np.where(np.minimum(((data[\"geoNetwork_country\"])), ((data[\"channelGrouping\"])))>0, ((data[\"channelGrouping\"]) - (1.0)), ((((-1.0) - (data[\"channelGrouping\"]))) / 2.0) )) \n    v[\"i95\"] = 0.067220*np.tanh((((np.minimum(((data[\"t_median\"])), ((np.where(data[\"sess_date_hours\"]>0, np.where(data[\"geoNetwork_city\"]>0, data[\"pred_0\"], data[\"t_median\"] ), 0.0 ))))) > (data[\"pred_0\"]))*1.)) \n    v[\"i96\"] = 0.100000*np.tanh(((((data[\"t_median\"]) * ((((((((data[\"geoNetwork_subContinent\"]) < ((((data[\"geoNetwork_subContinent\"]) < (data[\"geoNetwork_country\"]))*1.)))*1.)) * (data[\"geoNetwork_region\"]))) * 2.0)))) * 2.0)) \n    v[\"i97\"] = 0.099951*np.tanh(np.minimum((((((data[\"sess_date_dow\"]) > (np.tanh((data[\"geoNetwork_networkDomain\"]))))*1.))), ((np.where(data[\"geoNetwork_subContinent\"]<0, -3.0, (((data[\"totals_hits\"]) < (data[\"geoNetwork_city\"]))*1.) ))))) \n    v[\"i98\"] = 0.099951*np.tanh(np.minimum(((data[\"t_mean\"])), (((((((((2.0) - ((((data[\"t_mean\"]) < (data[\"totals_hits\"]))*1.)))) < (data[\"geoNetwork_city\"]))*1.)) * (-3.0)))))) \n    v[\"i99\"] = 0.075574*np.tanh(np.where((((data[\"geoNetwork_networkDomain\"]) > (data[\"geoNetwork_subContinent\"]))*1.)>0, data[\"geoNetwork_city\"], (((((data[\"geoNetwork_country\"]) * (data[\"geoNetwork_city\"]))) > (data[\"geoNetwork_subContinent\"]))*1.) ))\n\n    return Output(v.sum(axis=1))","execution_count":null,"outputs":[]},{"metadata":{"trusted":true,"_uuid":"008dc526d2eac28960e7dee6d73691453e83823b"},"cell_type":"code","source":"a = maxval*GP1(gp_trn_users)\nb = maxval*GP2(gp_trn_users)\nc = maxval*GP3(gp_trn_users)\nd = maxval*GP4(gp_trn_users)\ne = maxval*GP5(gp_trn_users)\nf = maxval*GP6(gp_trn_users)\nprint(mean_squared_error((maxval*gp_trn_users['target']), a) ** .5)\nprint(mean_squared_error((maxval*gp_trn_users['target']), b) ** .5)\nprint(mean_squared_error((maxval*gp_trn_users['target']), c) ** .5)\nprint(mean_squared_error((maxval*gp_trn_users['target']), d) ** .5)\nprint(mean_squared_error((maxval*gp_trn_users['target']), e) ** .5)\nprint(mean_squared_error((maxval*gp_trn_users['target']), f) ** .5)\nprint(mean_squared_error((maxval*gp_trn_users['target']), (a+b+c+d+e+f)/6) ** .5)","execution_count":null,"outputs":[]},{"metadata":{"trusted":true,"_uuid":"e254bfaa9fddc9207af644f1ddf19aa04a1e5b21"},"cell_type":"code","source":"a = maxval*GP1(gp_sub_users)\nb = maxval*GP2(gp_sub_users)\nc = maxval*GP3(gp_sub_users)\nd = maxval*GP4(gp_sub_users)\ne = maxval*GP5(gp_sub_users)\nf = maxval*GP6(gp_sub_users)","execution_count":null,"outputs":[]},{"metadata":{"trusted":true,"_uuid":"40eb7e55a8b1c0b24be2e431f993b1bc06c9df11"},"cell_type":"code","source":"gp_sub_users['PredictedLogRevenue'] = a.values \ngp_sub_users[['PredictedLogRevenue']].to_csv('gp1_test.csv', index=True) #1.4381\ngp_sub_users['PredictedLogRevenue'] = b.values\ngp_sub_users[['PredictedLogRevenue']].to_csv('gp2_test.csv', index=True) #1.4376\ngp_sub_users['PredictedLogRevenue'] = c.values\ngp_sub_users[['PredictedLogRevenue']].to_csv('gp3_test.csv', index=True) #1.4379\ngp_sub_users['PredictedLogRevenue'] = d.values\ngp_sub_users[['PredictedLogRevenue']].to_csv('gp4_test.csv', index=True) #1.4382\ngp_sub_users['PredictedLogRevenue'] = e.values\ngp_sub_users[['PredictedLogRevenue']].to_csv('gp5_test.csv', index=True) #1.4325\ngp_sub_users['PredictedLogRevenue'] = f.values\ngp_sub_users[['PredictedLogRevenue']].to_csv('gp6_test.csv', index=True) #1.4324\ngp_sub_users['PredictedLogRevenue'] = (a+b+c+d+e+f).values/6\ngp_sub_users[['PredictedLogRevenue']].to_csv('gpari_test.csv', index=True) ","execution_count":null,"outputs":[]}],"metadata":{"kernelspec":{"display_name":"Python 3","language":"python","name":"python3"},"language_info":{"name":"python","version":"3.6.6","mimetype":"text/x-python","codemirror_mode":{"name":"ipython","version":3},"pygments_lexer":"ipython3","nbconvert_exporter":"python","file_extension":".py"}},"nbformat":4,"nbformat_minor":1}