{"metadata":{"kernelspec":{"language":"python","display_name":"Python 3","name":"python3"},"language_info":{"name":"python","version":"3.12.12","mimetype":"text/x-python","codemirror_mode":{"name":"ipython","version":3},"pygments_lexer":"ipython3","nbconvert_exporter":"python","file_extension":".py"},"kaggle":{"accelerator":"nvidiaTeslaT4","dataSources":[{"sourceId":112509,"databundleVersionId":14254895,"sourceType":"competition"}],"dockerImageVersionId":31236,"isInternetEnabled":false,"language":"python","sourceType":"notebook","isGpuEnabled":true}},"nbformat_minor":4,"nbformat":4,"cells":[{"cell_type":"code","source":"# This Python 3 environment comes with many helpful analytics libraries installed\n# It is defined by the kaggle/python Docker image: https://github.com/kaggle/docker-python\n# For example, here's several helpful packages to load\n\nimport numpy as np # linear algebra\nimport pandas as pd # data processing, CSV file I/O (e.g. pd.read_csv)\n\n# Input data files are available in the read-only \"../input/\" directory\n# For example, running this (by clicking run or pressing Shift+Enter) will list all files under the input directory\n\nimport os\nfor dirname, _, filenames in os.walk('/kaggle/input'):\n    for filename in filenames:\n        print(os.path.join(dirname, filename))\n\n# You can write up to 20GB to the current directory (/kaggle/working/) that gets preserved as output when you create a version using \"Save & Run All\" \n# You can also write temporary files to /kaggle/temp/, but they won't be saved outside of the current session","metadata":{"_uuid":"8f2839f25d086af736a60e9eeb907d3b93b6e0e5","_cell_guid":"b1076dfc-b9ad-4769-8c92-a6c4dae69d19","trusted":true},"outputs":[],"execution_count":null},{"cell_type":"code","source":"","metadata":{"trusted":true},"outputs":[],"execution_count":null},{"cell_type":"code","source":"# ======================================\n# CELL 1 — KAGGLE SAFE TRAIN + SUBMISSION\n# ======================================\n\nimport os, warnings\nwarnings.filterwarnings(\"ignore\")\nos.environ[\"TF_CPP_MIN_LOG_LEVEL\"] = \"3\"\n\nimport numpy as np\nimport pandas as pd\nimport tensorflow as tf\nimport lightgbm as lgb\n\nfrom tensorflow.keras.preprocessing.image import load_img, img_to_array\nfrom tensorflow.keras.applications.efficientnet import EfficientNetB0, preprocess_input\n\n# ------------------\n# CONFIG\n# ------------------\nBASE_PATH = \"/kaggle/input/csiro-biomass\"\nTRAIN_CSV = f\"{BASE_PATH}/train.csv\"\nTEST_CSV  = f\"{BASE_PATH}/test.csv\"\n\nIMG_SIZE = 224\nBATCH = 16\nSEED = 42\n\nnp.random.seed(SEED)\ntf.random.set_seed(SEED)\n\nTARGETS = [\n    \"Dry_Green_g\",\n    \"Dry_Dead_g\",\n    \"Dry_Clover_g\",\n    \"GDM_g\",\n    \"Dry_Total_g\"\n]\n\n# ------------------\n# LOAD CSVs\n# ------------------\ndf_train = pd.read_csv(TRAIN_CSV)\ndf_test  = pd.read_csv(TEST_CSV)\n\nprint(f\"✅ Train rows: {len(df_train)} | Test rows: {len(df_test)}\")\n\n# ------------------\n# SAFE IMAGE LOADER (NO SKIPS EVER)\n# ------------------\ndef load_image_safe(path):\n    img = load_img(path, target_size=(IMG_SIZE, IMG_SIZE))\n    img = img_to_array(img)\n    return preprocess_input(img)\n\nX_img_train = np.zeros((len(df_train), IMG_SIZE, IMG_SIZE, 3), dtype=np.float32)\nX_img_test  = np.zeros((len(df_test),  IMG_SIZE, IMG_SIZE, 3), dtype=np.float32)\n\nfor i, p in enumerate(df_train[\"image_path\"]):\n    X_img_train[i] = load_image_safe(os.path.join(BASE_PATH, p))\n\nfor i, p in enumerate(df_test[\"image_path\"]):\n    X_img_test[i] = load_image_safe(os.path.join(BASE_PATH, p))\n\nprint(\"✅ Images loaded safely\")\n\n# ------------------\n# CNN FEATURE EXTRACTION (NO INTERNET)\n# ------------------\n# IMPORTANT:\n# weights=None avoids ANY download attempt\n# Kaggle kernels cannot fetch external URLs\n\ncnn = EfficientNetB0(\n    weights=None,              # 🔒 NO INTERNET\n    include_top=False,\n    pooling=\"avg\",\n    input_shape=(IMG_SIZE, IMG_SIZE, 3)\n)\n\nX_train = cnn.predict(X_img_train, batch_size=BATCH, verbose=0)\nX_test  = cnn.predict(X_img_test,  batch_size=BATCH, verbose=0)\n\nassert len(X_train) == len(df_train)\nassert len(X_test)  == len(df_test)\n\nprint(f\"✅ CNN features: {X_train.shape}\")\n\n# ------------------\n# LIGHTGBM TRAIN (FULL DATA)\n# ------------------\nparams = {\n    \"objective\": \"regression\",\n    \"learning_rate\": 0.03,\n    \"num_leaves\": 63,\n    \"feature_fraction\": 0.9,\n    \"bagging_fraction\": 0.8,\n    \"bagging_freq\": 5,\n    \"verbosity\": -1,\n    \"seed\": SEED\n}\n\npreds_test = np.zeros((len(df_test), len(TARGETS)))\n\nfor j, tgt in enumerate(TARGETS):\n    mask = df_train[\"target_name\"] == tgt\n    X_tgt = X_train[mask]\n    y_tgt = df_train.loc[mask, \"target\"].values\n\n    model = lgb.train(\n        params,\n        lgb.Dataset(X_tgt, y_tgt),\n        num_boost_round=800\n    )\n\n    preds_test[:, j] = model.predict(X_test)\n\npreds_test = np.clip(preds_test, 0, None)\n\n# ------------------\n# CREATE SUBMISSION (GUARANTEED)\n# ------------------\nrows = []\nfor i, sid in enumerate(df_test[\"sample_id\"]):\n    for j, tgt in enumerate(TARGETS):\n        rows.append({\n            \"sample_id\": f\"{sid}__{tgt}\",\n            \"target\": float(preds_test[i, j])\n        })\n\nsubmission = pd.DataFrame(rows)\nassert len(submission) == len(df_test) * 5\n\nsubmission.to_csv(\"submission.csv\", index=False)\nprint(\"✅ submission.csv written successfully\")\nsubmission.head()","metadata":{"trusted":true,"execution":{"iopub.status.busy":"2026-01-12T11:52:27.040856Z","iopub.execute_input":"2026-01-12T11:52:27.041362Z","iopub.status.idle":"2026-01-12T11:55:40.241598Z","shell.execute_reply.started":"2026-01-12T11:52:27.041334Z","shell.execute_reply":"2026-01-12T11:55:40.240627Z"}},"outputs":[],"execution_count":null},{"cell_type":"code","source":"# ======================================\n# CELL 2 — OFFLINE CV ONLY (OPTIONAL)\n# ======================================\n\nfrom sklearn.model_selection import GroupKFold\nfrom sklearn.metrics import r2_score\n\nweights = {\n    \"Dry_Green_g\":0.1,\n    \"Dry_Dead_g\":0.1,\n    \"Dry_Clover_g\":0.1,\n    \"GDM_g\":0.2,\n    \"Dry_Total_g\":0.5\n}\n\nscores = {}\n\nfor tgt in TARGETS:\n    mask = df_train[\"target_name\"] == tgt\n    X = X_train[mask]\n    y = df_train.loc[mask, \"target\"].values\n    groups = df_train.loc[mask, \"sample_id\"]\n\n    oof = np.zeros(len(y))\n    gkf = GroupKFold(n_splits=5)\n\n    for tr, va in gkf.split(X, y, groups):\n        model = lgb.train(\n            params,\n            lgb.Dataset(X[tr], y[tr]),\n            num_boost_round=500\n        )\n        oof[va] = model.predict(X[va])\n\n    scores[tgt] = r2_score(y, oof)\n    print(f\"✅ OOF R² ({tgt}): {scores[tgt]:.4f}\")\n\nweighted = sum(scores[t] * weights[t] for t in weights)\nprint(f\"\\n🌟 Weighted OOF R²: {weighted:.4f}\")","metadata":{"trusted":true,"execution":{"iopub.status.busy":"2026-01-12T12:10:00.631826Z","iopub.execute_input":"2026-01-12T12:10:00.632487Z","iopub.status.idle":"2026-01-12T12:11:07.813619Z","shell.execute_reply.started":"2026-01-12T12:10:00.632458Z","shell.execute_reply":"2026-01-12T12:11:07.812889Z"}},"outputs":[],"execution_count":null}]}