{"cells":[{"metadata":{"_uuid":"8f2839f25d086af736a60e9eeb907d3b93b6e0e5","_cell_guid":"b1076dfc-b9ad-4769-8c92-a6c4dae69d19","trusted":true},"cell_type":"code","source":"import numpy as np\nimport pandas as pd\nfrom zipfile import ZipFile\nimport matplotlib.pyplot as plt\nfrom scipy.signal import butter, lfilter, freqz\nfrom sklearn.metrics import f1_score\nfrom sklearn.model_selection import cross_val_score, cross_val_predict\nfrom sklearn.linear_model import LogisticRegression\nfrom sklearn.ensemble import RandomForestClassifier\nfrom sklearn.pipeline import make_pipeline\nfrom sklearn.preprocessing import StandardScaler\nfrom sklearn.svm import SVC\nfrom sklearn.metrics import accuracy_score, roc_auc_score, euclidean_distances\nfrom catboost import CatBoostClassifier\nfrom scipy.signal import decimate, freqz","execution_count":null,"outputs":[]},{"metadata":{"trusted":true},"cell_type":"code","source":"def get_subject_data_df(zip_file, subject_id, series_ids, folder):\n    df = pd.DataFrame()\n    for series_id in series_ids:\n        file_name = f'{folder}/subj{subject_id}_series{series_id}_data.csv'\n        unzipped_file = zip_file.open(file_name)\n        tmp_df = pd.read_csv(unzipped_file, index_col='id')\n        df = pd.concat([df, tmp_df])\n    return df\n\ndef get_subject_target_df(zip_file, subject_id):\n    df = pd.DataFrame()\n    for series_id in range(1,9):\n        file_name = f'train/subj{subject_id}_series{series_id}_events.csv'\n        unzipped_file = zip_file.open(file_name)\n        tmp_df = pd.read_csv(unzipped_file, index_col='id')\n        df = pd.concat([df, tmp_df])\n    return df","execution_count":null,"outputs":[]},{"metadata":{"trusted":true},"cell_type":"code","source":"def butter_lowpass(cutoff, fs, order=5):\n    nyq = 0.5 * fs\n    normal_cutoff = cutoff / nyq\n    b, a = butter(order, normal_cutoff, btype='low', analog=False)\n    return b, a\n\ndef butter_lowpass_filter(data, cutoff, fs, order=5):\n    b, a = butter_lowpass(cutoff, fs, order=order)\n    freqz(b,a)\n    y = lfilter(b, a, data, axis=0)\n    return y","execution_count":null,"outputs":[]},{"metadata":{"trusted":true},"cell_type":"code","source":"def butter_highpass(cutoff, fs, order=5):\n    nyq = 0.5 * fs\n    normal_cutoff = cutoff / nyq\n    b, a = butter(order, normal_cutoff, btype='high', analog=False)\n    return b, a\n\ndef butter_highpass_filter(data, cutoff, fs, order=5):\n    b, a = butter_highpass(cutoff, fs, order=order)\n    y = lfilter(b, a, data, axis=0)\n    return y","execution_count":null,"outputs":[]},{"metadata":{"trusted":true},"cell_type":"code","source":"def preprocess_data(x, downsample_rate):    \n    signal = decimate(x, downsample_rate, axis=0)\n    return signal","execution_count":null,"outputs":[]},{"metadata":{"trusted":true,"scrolled":true},"cell_type":"code","source":"subject_ids = range(1,13)\n# subject_ids = subject_ids[:1]\ntrain_zip_path = '/kaggle/input/grasp-and-lift-eeg-detection/train.zip'\ntest_zip_path = '/kaggle/input/grasp-and-lift-eeg-detection/test.zip'\ntrain_zip = ZipFile(train_zip_path)\ntest_zip = ZipFile(test_zip_path)\n\nsubmission = []\nidx = []\ndroprate = 50\nfs = 500\nlow_cut_off = 0.2\nhigh_cut_off = 4\nauc = []\n\nfor subject_id in subject_ids:\n    x_train = get_subject_data_df(train_zip, subject_id, range(1, 9), 'train')\n    x_test = get_subject_data_df(test_zip, subject_id, [9,10], 'test')\n    target_df = get_subject_target_df(train_zip, subject_id)\n    idx.append(np.array(x_test.index))\n\n    m,n = x_test.shape\n    pred = np.zeros((m, 6))\n    scaler = StandardScaler()\n    x_train = butter_lowpass_filter(x_train, low_cut_off, fs)\n    x_train = butter_highpass_filter(x_train, high_cut_off, fs)\n    scaler.fit(x_train)    \n    x = x_train[::droprate]\n    x = scaler.transform(x)\n    x_test = butter_lowpass_filter(x_test, low_cut_off, fs)\n    x_test = butter_highpass_filter(x_test, high_cut_off, fs)\n    x_test = scaler.transform(x_test)\n    \n    cb = LogisticRegression(solver='lbfgs', max_iter=200)\n    \n    for i in range(6):\n        y = target_df[::droprate].values[:, i]\n        score = cross_val_score(cb, x, y, cv=2, scoring='roc_auc').mean()\n        auc.append(score)\n        cb.fit(x,y)\n        pred[:,i] = cb.predict_proba(x_test)[:, 1]\n    submission.append(pred)\n    print('end of subject: ', subject_id)","execution_count":null,"outputs":[]},{"metadata":{"trusted":true},"cell_type":"code","source":"auc","execution_count":null,"outputs":[]},{"metadata":{"trusted":true},"cell_type":"code","source":"np.mean(auc)","execution_count":null,"outputs":[]},{"metadata":{"_kg_hide-output":false,"trusted":true},"cell_type":"code","source":"submission_concat = np.concatenate(submission, axis=0)\nidx_concat = np.concatenate(idx)","execution_count":null,"outputs":[]},{"metadata":{"trusted":true},"cell_type":"code","source":"cols = ['HandStart','FirstDigitTouch',\n        'BothStartLoadPhase','LiftOff',\n        'Replace','BothReleased']\nsubmission_csv = pd.DataFrame(index=idx_concat, columns=cols, data=submission_concat)","execution_count":null,"outputs":[]},{"metadata":{"trusted":true},"cell_type":"code","source":"submission_csv.to_csv('submission_classes.csv.gz', index_label='id', float_format='%.5f',  compression='gzip')","execution_count":null,"outputs":[]},{"metadata":{},"cell_type":"markdown","source":"1. Пока это только норм работает ниже всякая херня для дальнейшей работы"},{"metadata":{},"cell_type":"markdown","source":"Test"},{"metadata":{},"cell_type":"markdown","source":"# Singular values aproach"},{"metadata":{"trusted":true},"cell_type":"code","source":"# import torch.nn as nn\n# import torch.nn.functional as F","execution_count":null,"outputs":[]},{"metadata":{"trusted":true},"cell_type":"code","source":"# class CNN(nn.Module):\n#     def __init__(self):\n#         super(CNN, self).__init__()\n#         self.conv1 = nn.Conv2d(3, 6, 5)\n#         self.pool = nn.MaxPool2d(2, 2)\n#         self.conv2 = nn.Conv2d(6, 16, 5)\n#         self.fc1 = nn.Linear(16 * 5 * 5, 120)\n#         self.fc2 = nn.Linear(120, 84)\n#         self.fc3 = nn.Linear(84, 6)\n\n#     def forward(self, x):\n#         x = self.pool(F.relu(self.conv1(x)))\n#         x = self.pool(F.relu(self.conv2(x)))\n#         x = x.view(-1, 16 * 5 * 5)\n#         x = F.relu(self.fc1(x))\n#         x = F.relu(self.fc2(x))\n#         x = self.fc3(x)\n#         x = F.sigmoid(x)\n#         return x","execution_count":null,"outputs":[]},{"metadata":{"trusted":true},"cell_type":"code","source":"# net = CNN()","execution_count":null,"outputs":[]},{"metadata":{"trusted":true},"cell_type":"code","source":"# torch_singular_values = torch.Tensor(singular_values)\n# torch_mean_targets = torch.Tensor(mean_targets)\n\n# optimizer = optim.Adam(net.parameters())\n# loss_function = nn.BCEWithLogitsLoss()\n# batch_size = 32\n# batch_count = batches // batch_size\n\n# output = net.forward(torch_singular_values)\n# loss = loss_function(output, torch_mean_targets)\n# print(loss)\n\n# net.train()\n# for epoch in range(40):\n#     for i in range(batch_count):\n#         batch = torch_singular_values[i * batch_size: (i + 1) * batch_size]\n#         target = torch_mean_targets[i * batch_size: (i + 1) * batch_size]\n#         optimizer.zero_grad()\n#         output = net.forward(batch)\n#         loss = loss_function(output, target)    \n#         loss.backward()\n\n#         optimizer.step()\n\n#         output = net.forward(torch_singular_values)\n#         loss = loss_function(output, torch_mean_targets)\n#         print(loss.data)","execution_count":null,"outputs":[]},{"metadata":{"trusted":true},"cell_type":"code","source":"# window_size = 1000\n# n,m = data_df.shape\n# batches = n // window_size + 1\n# singular_values = np.zeros((batches, m))\n# for i in range(0, batches):\n#     window = data_df[i * window_size : (i + 1) * window_size]\n#     u,s,vh = np.linalg.svd(window, False)\n#     singular_values[i] = s","execution_count":null,"outputs":[]},{"metadata":{"trusted":true},"cell_type":"code","source":"# mean_targets = np.zeros((batches, 6))\n# for i in range(0, batches):\n#     target_window = target_df[i * window_size : (i + 1) * window_size]\n#     mean_target = np.mean(target_window)     \n#     mean_targets[i] = mean_target","execution_count":null,"outputs":[]},{"metadata":{"trusted":true},"cell_type":"code","source":"# import torch\n# import torch.nn as nn\n# import torch.nn.functional as F\n# import torch.optim as optim","execution_count":null,"outputs":[]},{"metadata":{"trusted":true},"cell_type":"code","source":"# class Net(nn.Module):\n#     def __init__(self, in_size):\n#         super(Net, self).__init__()\n#         self.lin_layer1 = nn.Linear(in_size, 100)\n#         self.lin_layer2 = nn.Linear(100, 100)\n#         self.lin_layer3 = nn.Linear(100, 6)\n#         self.Act = nn.ReLU()\n#         self.Sigm = nn.Sigmoid()\n\n#     def forward(self, x):\n#         x = self.lin_layer1(x)\n#         x = self.Act(x)\n#         x = self.lin_layer2(x)\n#         x = self.Act(x)\n#         x = self.lin_layer3(x)\n#         # x = self.Sigm(x)\n#         return x","execution_count":null,"outputs":[]},{"metadata":{"trusted":true},"cell_type":"code","source":"# m, n = singular_values.shape\n# net = Net(n)","execution_count":null,"outputs":[]},{"metadata":{"trusted":true},"cell_type":"code","source":"# torch_singular_values = torch.Tensor(singular_values)\n# torch_mean_targets = torch.Tensor(mean_targets)\n\n# optimizer = optim.Adam(net.parameters())\n# loss_function = nn.BCEWithLogitsLoss()\n# batch_size = 32\n# batch_count = batches // batch_size\n\n# output = net.forward(torch_singular_values)\n# loss = loss_function(output, torch_mean_targets)\n# print(loss)\n\n# net.train()\n# for epoch in range(40):\n#     for i in range(batch_count):\n#         batch = torch_singular_values[i * batch_size: (i + 1) * batch_size]\n#         target = torch_mean_targets[i * batch_size: (i + 1) * batch_size]\n#         optimizer.zero_grad()\n#         output = net.forward(batch)\n#         loss = loss_function(output, target)    \n#         loss.backward()\n\n#         optimizer.step()\n\n#         output = net.forward(torch_singular_values)\n#         loss = loss_function(output, torch_mean_targets)\n#         print(loss.data)\n    ","execution_count":null,"outputs":[]},{"metadata":{"trusted":true},"cell_type":"code","source":"# def get_subject_test_data_df(zip_file, subject_id):\n#     df = pd.DataFrame()\n#     for series_id in range(9, 11):\n#         file_name = f'test/subj{subject_id}_series{series_id}_data.csv'\n#         unzipped_file = zip_file.open(file_name)\n#         tmp_df = pd.read_csv(unzipped_file, index_col='id')\n#         df = pd.concat([df, tmp_df])\n#     return df","execution_count":null,"outputs":[]},{"metadata":{"trusted":true},"cell_type":"code","source":"# test_zip_path = '/kaggle/input/grasp-and-lift-eeg-detection/test.zip'\n# test_zip = ZipFile(test_zip_path)\n# for subject_id in subject_ids:\n#     test_data_df = get_subject_test_data_df(test_zip, subject_id)","execution_count":null,"outputs":[]},{"metadata":{"trusted":true},"cell_type":"code","source":"# n,m = test_data_df.shape\n# batches = n // window_size + 1\n# test_singular_values = np.zeros((batches, m))\n# for i in range(0, batches):\n#     window = test_data_df[i * window_size : (i + 1) * window_size]\n#     u,s,vh = np.linalg.svd(window, False)\n#     test_singular_values[i] = s","execution_count":null,"outputs":[]},{"metadata":{"trusted":true},"cell_type":"code","source":"# torch_test_singular_values = torch.Tensor(test_singular_values)\n# test_output = net.forward(torch_test_singular_values)\n# test_output = nn.functional.sigmoid(test_output)","execution_count":null,"outputs":[]},{"metadata":{"trusted":true},"cell_type":"code","source":"# sample_submission_path = '/kaggle/input/grasp-and-lift-eeg-detection/sample_submission.csv.zip'\n# submission = pd.read_csv(sample_submission_path, index_col='id')","execution_count":null,"outputs":[]},{"metadata":{"trusted":true},"cell_type":"code","source":"# out = np.array(test_output.data)\n# out = out.repeat(1000, axis=0)\n# m,n = out.shape\n# submission.iloc[:m, :] = out","execution_count":null,"outputs":[]},{"metadata":{"trusted":true},"cell_type":"code","source":"# submission.to_csv('result2.csv')","execution_count":null,"outputs":[]}],"metadata":{"kernelspec":{"language":"python","display_name":"Python 3","name":"python3"},"language_info":{"pygments_lexer":"ipython3","nbconvert_exporter":"python","version":"3.6.4","file_extension":".py","codemirror_mode":{"name":"ipython","version":3},"name":"python","mimetype":"text/x-python"}},"nbformat":4,"nbformat_minor":1}