{"cells":[{"metadata":{"_uuid":"8f2839f25d086af736a60e9eeb907d3b93b6e0e5","_cell_guid":"b1076dfc-b9ad-4769-8c92-a6c4dae69d19","trusted":true,"collapsed":true},"cell_type":"code","source":"# I'm just taking these three results and naively mixing them together through different kinds of means\nimport pandas as pd\nimport numpy as np","execution_count":3,"outputs":[]},{"metadata":{"_uuid":"d629ff2d2480ee46fbb7e2d37f6b5fab8052498a","_cell_guid":"79c7e3d0-c299-4dcb-8224-4455121ee9b0","trusted":true},"cell_type":"code","source":"print(\"Reading the data...\\n\")\ndf1 = pd.read_csv('../input/r-lgbm-single-model-40m-rows-lb-0-9736/lgb_Usrnewness.csv')\ndf2 = pd.read_csv('../input/simple-linear-stacking-with-ranks-lb-0-9760/sub_stacked.csv')\ndf3 = pd.read_csv('../input/talkingdata-wordbatch-fm-ftrl-lb-0-9752/wordbatch_fm_ftrl.csv')","execution_count":4,"outputs":[]},{"metadata":{"_uuid":"aefcee01fe5be8e875c8cde837d8db9f8c28d001","_cell_guid":"402a543d-4c27-4a7f-80ff-16bca23e1a02","trusted":true,"collapsed":true},"cell_type":"code","source":"models = { 'df1' : {\n                    'name':'r-lgbm',\n                    'score':97.36,\n                    'df':df1 },\n          'df2' : {\n                    'name':'linstack',\n                    'score':97.60,\n                    'df':df2 },\n           'df3' : {\n                    'name':'wordbatch',\n                    'score':97.52,\n                    'df':df3 }\n         }","execution_count":5,"outputs":[]},{"metadata":{"_uuid":"40edd5db64b45ceb1989b3af6653381e44b6ca51","_cell_guid":"c7d67382-8f2e-406c-b0da-ed4e83163178","trusted":true},"cell_type":"code","source":"df1.head()","execution_count":6,"outputs":[]},{"metadata":{"_uuid":"5a5ceb3704cde86e3fa88be2398d8bb80ec6e243","_cell_guid":"cb911e40-6d42-47d3-8ee0-f8e4d64cd03e","trusted":true},"cell_type":"code","source":"# Making simple blendings of the models\n\nisa_lg = 0\nisa_hm = 0\nprint(\"Blending...\\n\")\nfor df in models.keys() : \n    isa_lg += np.log(models[df]['df'].is_attributed)\n    isa_hm += 1/(models[df]['df'].is_attributed)\nisa_lg = np.exp(isa_lg/3)\nisa_hm = 1/isa_hm\n\nprint(\"Isa log\\n\")\nprint(isa_lg[:5])\nprint()\nprint(\"Isa harmo\\n\")\nprint(isa_hm[:5])","execution_count":7,"outputs":[]},{"metadata":{"_uuid":"119edd431db392fe065c18018fe3fd7fe3debffc","_cell_guid":"23562114-bd76-4b5c-9244-09e7d31647cd","trusted":true},"cell_type":"code","source":"sub_log = pd.DataFrame()\nsub_log['click_id'] = df1['click_id']\nsub_log['is_attributed'] = isa_lg\nsub_log.head()","execution_count":8,"outputs":[]},{"metadata":{"_uuid":"a57979a742f7e6f2be2078a8170eabd4452ad4a3","_cell_guid":"dd46ee67-c5be-4b36-a315-1c8279c496e1","trusted":true},"cell_type":"code","source":"sub_hm = pd.DataFrame()\nsub_hm['click_id'] = df1['click_id']\nsub_hm['is_attributed'] = isa_hm\nsub_hm.head()","execution_count":null,"outputs":[]},{"metadata":{"_uuid":"7500358c04e8750f48565f45238dc92ed7445443","_cell_guid":"68aba841-020e-42fc-9465-4cfb380f20c3","trusted":true},"cell_type":"code","source":"print(\"Writing...\")\nsub_log.to_csv('sub_log.csv', index=False, float_format='%.9f')\nsub_hm.to_csv('sub_hm.csv', index=False, float_format='%.9f')","execution_count":null,"outputs":[]}],"metadata":{"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"},"kernelspec":{"display_name":"Python 3","language":"python","name":"python3"}},"nbformat":4,"nbformat_minor":1}