{"cells":[{"metadata":{"_uuid":"8f2839f25d086af736a60e9eeb907d3b93b6e0e5","_cell_guid":"b1076dfc-b9ad-4769-8c92-a6c4dae69d19","trusted":true,"_kg_hide-input":false},"cell_type":"code","source":"import seaborn as sns\nimport matplotlib.pyplot as plt\nimport numpy as np \nimport pandas as pd \nimport os\nimport PIL\n\nfrom PIL import ImageOps\nfrom tqdm import tqdm\n\ntqdm.pandas()","execution_count":1,"outputs":[]},{"metadata":{"trusted":true,"_kg_hide-input":false},"cell_type":"code","source":"df_train = pd.read_csv('../input/train.csv')\nprint('Train.csv loaded!')","execution_count":2,"outputs":[{"output_type":"stream","text":"Train.csv loaded!\n","name":"stdout"}]},{"metadata":{"trusted":true,"_kg_hide-input":false},"cell_type":"code","source":"def get_img_size(img_id, path): \n img = PIL.Image.open(f'{path}{img_id}.png')\n return img.size","execution_count":3,"outputs":[]},{"metadata":{"trusted":true,"_kg_hide-input":false},"cell_type":"code","source":"train_path = '../input/train/'\ndf_train['shape'] = df_train['id'].progress_apply(lambda x : get_img_size(x, train_path))\ndf_train['width'] = [width for width,_ in df_train['shape'].values]\ndf_train['height'] = [height for _,height in df_train['shape'].values]\ndf_train['aspect_ratio'] = df_train.height/df_train.width\ndf_train['std_width'] = (df_train.width - np.mean(df_train.width)) / np.std(df_train.width)\ndf_train['std_height'] = (df_train.height - np.mean(df_train.height)) / np.std(df_train.height)\ndf_train['std_aspect_ratio'] = (df_train.aspect_ratio - np.mean(df_train.aspect_ratio)) / np.std(df_train.aspect_ratio)\ndf_train.to_csv('train_with_wh.csv', index=False)","execution_count":4,"outputs":[{"output_type":"stream","text":"100%|██████████| 109237/109237 [04:38<00:00, 392.91it/s]\n","name":"stderr"}]},{"metadata":{"trusted":true},"cell_type":"code","source":"df_train.head()","execution_count":5,"outputs":[{"output_type":"execute_result","execution_count":5,"data":{"text/plain":" id ... std_aspect_ratio\n0 1000483014d91860 ... -0.462639\n1 1000fe2e667721fe ... -0.796301\n2 1001614cb89646ee ... -0.582326\n3 10041eb49b297c08 ... 0.122864\n4 100501c227f8beea ... 1.198770\n\n[5 rows x 9 columns]","text/html":"
\n\n\n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n
idattribute_idsshapewidthheightaspect_ratiostd_widthstd_heightstd_aspect_ratio
01000483014d91860147 616 813(339, 300)3393000.884956-0.179729-0.582173-0.462639
11000fe2e667721fe51 616 734 813(423, 300)4233000.7092200.337904-0.582173-0.796301
21001614cb89646ee776(365, 300)3653000.821918-0.019509-0.582173-0.582326
310041eb49b297c0851 671 698 813 1092(300, 358)3003581.193333-0.420058-0.1330530.122864
4100501c227f8beea13 404 492 903 1093(300, 528)3005281.760000-0.4200581.1833331.198770
\n
"},"metadata":{}}]},{"metadata":{"trusted":true},"cell_type":"code","source":"df_test = pd.read_csv('../input/sample_submission.csv')","execution_count":6,"outputs":[]},{"metadata":{"trusted":true},"cell_type":"code","source":"test_path = '../input/test/'\ndf_test['shape'] = df_test['id'].progress_apply(lambda x : get_img_size(x, test_path))\ndf_test['width'] = [width for width,_ in df_test['shape'].values]\ndf_test['height'] = [height for _,height in df_test['shape'].values]\ndf_test['aspect_ratio'] = df_test.height/df_test.width\ndf_test['std_width'] = (df_test.width - np.mean(df_test.width)) / np.std(df_test.width)\ndf_test['std_height'] = (df_test.height - np.mean(df_test.height)) / np.std(df_test.height)\ndf_test['std_aspect_ratio'] = (df_test.aspect_ratio - np.mean(df_test.aspect_ratio)) / np.std(df_test.aspect_ratio)\ndf_test.to_csv('test_with_wh.csv', index=False)","execution_count":7,"outputs":[{"output_type":"stream","text":"100%|██████████| 7443/7443 [00:17<00:00, 437.22it/s]\n","name":"stderr"}]},{"metadata":{"trusted":true},"cell_type":"code","source":"df_test.head()","execution_count":8,"outputs":[{"output_type":"execute_result","execution_count":8,"data":{"text/plain":" id ... std_aspect_ratio\n0 10023b2cc4ed5f68 ... 1.321055\n1 100fbe75ed8fd887 ... -1.480735\n2 101b627524a04f19 ... -0.740811\n3 10234480c41284c6 ... -0.617821\n4 1023b0e2636dcea8 ... -0.737010\n\n[5 rows x 9 columns]","text/html":"
\n\n\n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n
idattribute_idsshapewidthheightaspect_ratiostd_widthstd_heightstd_aspect_ratio
010023b2cc4ed5f680 1 2(300, 533)3005331.776667-0.4103981.3484111.321055
1100fbe75ed8fd8870 1 2(781, 300)7813000.3841232.499366-0.602236-1.480735
2101b627524a04f190 1 2(399, 300)3993000.7518800.188493-0.602236-0.740811
310234480c41284c60 1 2(369, 300)3693000.8130080.007011-0.602236-0.617821
41023b0e2636dcea80 1 2(398, 300)3983000.7537690.182444-0.602236-0.737010
\n
"},"metadata":{}}]}],"metadata":{"kernelspec":{"display_name":"Python 3","language":"python","name":"python3"},"language_info":{"name":"python","version":"3.6.4","mimetype":"text/x-python","codemirror_mode":{"name":"ipython","version":3},"pygments_lexer":"ipython3","nbconvert_exporter":"python","file_extension":".py"}},"nbformat":4,"nbformat_minor":1}