{"cells":[{"metadata":{"trusted":true},"cell_type":"code","source":"'''\nsubmit of only B3 B4 & B5 models\n'''","execution_count":null,"outputs":[]},{"metadata":{"_uuid":"8f2839f25d086af736a60e9eeb907d3b93b6e0e5","_cell_guid":"b1076dfc-b9ad-4769-8c92-a6c4dae69d19","trusted":true},"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\n#for 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 5GB 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","execution_count":null,"outputs":[]},{"metadata":{"_uuid":"d629ff2d2480ee46fbb7e2d37f6b5fab8052498a","_cell_guid":"79c7e3d0-c299-4dcb-8224-4455121ee9b0","trusted":true},"cell_type":"code","source":"# size 384 : All Preds from EB\n#pred_b2 = pd.read_csv('../input/rcsiimpreds/sub_EfficientNetB2_384.csv') -- it is decreasing LB score\npred_b3 = pd.read_csv('../input/rcsiimpreds/sub_EfficientNetB3_384_9460.csv') \npred_b4 = pd.read_csv('../input/rcsiimpreds/sub_EfficientNetB4_384_9498.csv') \npred_b5 = pd.read_csv('../input/rcsiimpreds/sub_EfficientNetB5_384_9454.csv') \npred_b6 = pd.read_csv('../input/rcsiimpreds/sub_EfficientNetB6_384_9481.csv')\n# class weight\npred_cw_b4 = pd.read_csv('../input/rcsiimpreds/sub_EfficientNetB4_CW_384_9457.csv')\npred_cw_b4.rename(columns = {'target':'target_cw_b4'}, inplace = True)","execution_count":null,"outputs":[]},{"metadata":{"trusted":true},"cell_type":"code","source":"# Prediction from B5 run one by one for each fold\npred_512_B6 = pd.read_csv('../input/rcsiimpreds/sub_B5_512_3fold_9466.csv')\npred_512_B6.rename(columns = {'target':'target_B6_512'}, inplace = True)","execution_count":null,"outputs":[]},{"metadata":{"trusted":true},"cell_type":"code","source":"pred_tta_b3 = pd.read_csv('../input/rcsiimpreds/siim_tta_b3_9458.csv')\npred_tta_b3.rename(columns = {'target':'target_tta_b3'}, inplace = True)\n\npred_tta_b4 = pd.read_csv('../input/rcsiimpreds/siim_tta_b4_9473.csv')\npred_tta_b4.rename(columns = {'target':'target_tta_b4'}, inplace = True)\n\nresult_tta = pd.merge(pred_b3, pred_b4, on='image_name',suffixes=('_tta_b3','_tta_b4'))\n\nresult_tta.head()","execution_count":null,"outputs":[]},{"metadata":{"trusted":true},"cell_type":"code","source":"# B4 on 512\n#pred_512_b4   = pd.read_csv('../input/rcsiimpreds/sub_EfficientNetB4_512_9448.csv')","execution_count":null,"outputs":[]},{"metadata":{"trusted":true},"cell_type":"code","source":"#pred_512_B5_F3.rename(columns = {'target':'target_F3'}, inplace = True) ","execution_count":null,"outputs":[]},{"metadata":{"trusted":true},"cell_type":"code","source":"result1 = pd.merge(pred_b3, pred_b4, on='image_name',suffixes=('_b3','_b4'))","execution_count":null,"outputs":[]},{"metadata":{"trusted":true},"cell_type":"code","source":"result1.head()","execution_count":null,"outputs":[]},{"metadata":{"trusted":true},"cell_type":"code","source":"result2 = pd.merge(pred_b5, pred_b6, on='image_name',suffixes=('_b5','_b6'))","execution_count":null,"outputs":[]},{"metadata":{"trusted":true},"cell_type":"code","source":"result2.head()","execution_count":null,"outputs":[]},{"metadata":{"trusted":true},"cell_type":"code","source":"semi_final = pd.merge(result1, result2, on='image_name')","execution_count":null,"outputs":[]},{"metadata":{"trusted":true},"cell_type":"code","source":"semi_final.head()","execution_count":null,"outputs":[]},{"metadata":{"trusted":true},"cell_type":"code","source":"result3 = pd.merge(pred_cw_b4, pred_512_B6, on='image_name',suffixes=('_cw_b4','_B6_512'))\nresult3.head()","execution_count":null,"outputs":[]},{"metadata":{"trusted":true},"cell_type":"code","source":"final = pd.merge(semi_final, result3, on='image_name')\nfinal.head()","execution_count":null,"outputs":[]},{"metadata":{"trusted":true},"cell_type":"code","source":"final = pd.merge(final, result_tta, on='image_name')\nfinal.head()","execution_count":null,"outputs":[]},{"metadata":{"trusted":true},"cell_type":"code","source":"# Adding Karan's model prediction\npred_kr_b3 = pd.read_csv('../input/rcsiimpreds/KR_sub_EfficientNetB3_512_9520.csv')\npred_kr_b3.rename(columns = {'target':'target_kr_b3'}, inplace = True)\n\npred_kr_b4 = pd.read_csv('../input/rcsiimpreds/KR_sub_EfficientNetB4_512_9499.csv')\npred_kr_b4.rename(columns = {'target':'target_kr_b4'}, inplace = True)\n\npred_kr_eb3 = pd.read_csv('../input/rcsiimpreds/KR_sub_eb3_512_9554.csv')\npred_kr_eb3.rename(columns = {'target':'target_kr_eb3'}, inplace = True)\n\n#pred_cw_b4 = pd.read_csv('')\n#pred_cw_b4.rename(columns = {'target':'target_cw_b4'}, inplace = True)\n","execution_count":null,"outputs":[]},{"metadata":{"trusted":true},"cell_type":"code","source":"kr_result = pd.merge(pred_kr_b3, pred_kr_b4, on='image_name',suffixes=('_b3','_b4'))\n\nkr_result = pd.merge(kr_result, pred_kr_eb3, on='image_name',suffixes=('_b3','_b4'))\nkr_result.head()","execution_count":null,"outputs":[]},{"metadata":{"trusted":true},"cell_type":"code","source":"final = pd.merge(final, kr_result, on='image_name')\nfinal.head()","execution_count":null,"outputs":[]},{"metadata":{"trusted":true},"cell_type":"code","source":"pred_256_b4 = pd.read_csv('../input/rcsiimpreds/sub_EfficientNetB4_256_9496.csv')\npred_256_b4.rename(columns = {'target':'target_256_b4'}, inplace = True)\npred_256_b4.head()","execution_count":null,"outputs":[]},{"metadata":{"trusted":true},"cell_type":"code","source":"final = pd.merge(final, pred_256_b4, on='image_name')\nfinal.head()","execution_count":null,"outputs":[]},{"metadata":{"trusted":true},"cell_type":"code","source":"# Removing low LB score prediction\n# target_tta_b3, target_b5, target_cw_b4\n#\nfinal['target'] = ( (final.target_b4)  + \n                   (final.target_b6)  + \n                   final.target_tta_b4 +\n                   final.target_kr_b3 + final.target_kr_b4 + final.target_kr_eb3 +\n                   final.target_256_b4\n                  ) / 7\nfinal.head()","execution_count":null,"outputs":[]},{"metadata":{"trusted":true},"cell_type":"code","source":"from scipy.stats import gmean","execution_count":null,"outputs":[]},{"metadata":{"trusted":true},"cell_type":"code","source":"# Removing low LB score prediction\n# target_tta_b3, target_b5, target_cw_b4\n#\ntarget_array =  np.array([final.target_b4,    final.target_b6   ,  final.target_tta_b4, final.target_B6_512,\n                       final.target_kr_b3, final.target_kr_b4,  final.target_kr_eb3, final.target_256_b4])\nfinal['target'] = gmean(target_array)\nfinal.head()","execution_count":null,"outputs":[]},{"metadata":{"trusted":true},"cell_type":"code","source":"final.head()","execution_count":null,"outputs":[]},{"metadata":{"trusted":true},"cell_type":"code","source":"submit_file = final[['image_name', 'target']]","execution_count":null,"outputs":[]},{"metadata":{"trusted":true},"cell_type":"code","source":"submit_file.head()","execution_count":null,"outputs":[]},{"metadata":{"trusted":true},"cell_type":"code","source":"submit_file.to_csv('submission.csv', index = False)","execution_count":null,"outputs":[]},{"metadata":{"trusted":true},"cell_type":"code","source":"submit_file.target.hist()","execution_count":null,"outputs":[]},{"metadata":{"trusted":true},"cell_type":"markdown","source":"# Min MAX\n\nimport numpy as np\nimport pandas as pd \nimport os \n\n# list file here to remove\nto_remove = ['sub_EfficientNetB2_384.csv',\n             'sub_EfficientNetB4_512_9448.csv',\n             'sub_EfficientNetB3_512_9403.csv']\n\ndef MinMaxBestBaseStacking(input_folder, best_base, output_path):\n    sub_base = pd.read_csv(best_base)\n    all_files = os.listdir(input_folder)\n    \n    all_files.remove(to_remove[0])\n    all_files.remove(to_remove[1])\n    all_files.remove(to_remove[2])\n        \n\n    # Read and concatenate submissions\n    outs = [pd.read_csv(os.path.join(input_folder, f), index_col=0) for f in all_files]\n    concat_sub = pd.concat(outs, axis=1)\n    cols = list(map(lambda x: \"target\" + str(x), range(len(concat_sub.columns))))\n    concat_sub.columns = cols\n    concat_sub.reset_index(inplace=True)\n    print(concat_sub.head())\n\n    # get the data fields ready for stacking\n    col_number = len(all_files) + 1\n    concat_sub['is_iceberg_max'] = concat_sub.iloc[:, 1:col_number].max(axis=1)\n    concat_sub['is_iceberg_min'] = concat_sub.iloc[:, 1:col_number].min(axis=1)\n    concat_sub['is_iceberg_mean'] = concat_sub.iloc[:,1:col_number].mean(axis=1)\n    concat_sub['is_iceberg_median'] = concat_sub.iloc[:, 1:col_number].median(axis=1)\n\n    # set up cutoff threshold for lower and upper bounds\n    cutoff_lo = 0.73\n    cutoff_hi = 0.33\n\n    concat_sub['is_iceberg_base'] = sub_base['target']\n    concat_sub['target'] = np.where(np.all(concat_sub.iloc[:, 1:col_number] > cutoff_lo, axis=1),\n                                        concat_sub['is_iceberg_max'],\n                                        np.where(np.all(concat_sub.iloc[:, 1:col_number] < cutoff_hi, axis=1),\n                                                 concat_sub['is_iceberg_min'],\n                                                 concat_sub['is_iceberg_base']))\n    concat_sub[['image_name', 'target']].to_csv(output_path,\n                                            index=False, float_format='%.12f')\n","execution_count":null},{"metadata":{"trusted":true},"cell_type":"markdown","source":"MinMaxBestBaseStacking('../input/rcsiimpreds/', \n                       '../input/rcsiimpreds/sub_EfficientNetB4_384_9498.csv', \n                       'submission_minmax.csv') # 0.9526 ","execution_count":null},{"metadata":{"trusted":true},"cell_type":"code","source":"#all_files = os.listdir('../input/rcsiimpreds/')\n#print(all_files)","execution_count":null,"outputs":[]},{"metadata":{"trusted":true},"cell_type":"code","source":"","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":4}