{"id":13937314,"url":"https://github.com/DmitryKey/kaggle_stackexchange_prediction","last_synced_at":"2025-07-19T23:31:55.052Z","repository":{"id":71450728,"uuid":"13764604","full_name":"DmitryKey/kaggle_stackexchange_prediction","owner":"DmitryKey","description":"This is source code of submission by SkyNet team for the StackExchange contest of closed question prediction.","archived":false,"fork":false,"pushed_at":"2018-03-17T11:10:26.000Z","size":50929,"stargazers_count":1,"open_issues_count":0,"forks_count":0,"subscribers_count":3,"default_branch":"master","last_synced_at":"2024-11-27T05:32:30.000Z","etag":null,"topics":[],"latest_commit_sha":null,"homepage":null,"language":"Python","has_issues":true,"has_wiki":null,"has_pages":null,"mirror_url":null,"source_name":null,"license":null,"status":null,"scm":"git","pull_requests_enabled":true,"icon_url":"https://github.com/DmitryKey.png","metadata":{"files":{"readme":"README.md","changelog":null,"contributing":null,"funding":null,"license":null,"code_of_conduct":null,"threat_model":null,"audit":null,"citation":null,"codeowners":null,"security":null,"support":null,"governance":null,"roadmap":null,"authors":null,"dei":null,"publiccode":null,"codemeta":null}},"created_at":"2013-10-22T06:09:40.000Z","updated_at":"2014-07-14T10:55:27.000Z","dependencies_parsed_at":"2023-03-01T12:15:42.932Z","dependency_job_id":null,"html_url":"https://github.com/DmitryKey/kaggle_stackexchange_prediction","commit_stats":null,"previous_names":[],"tags_count":0,"template":false,"template_full_name":null,"purl":"pkg:github/DmitryKey/kaggle_stackexchange_prediction","repository_url":"https://repos.ecosyste.ms/api/v1/hosts/GitHub/repositories/DmitryKey%2Fkaggle_stackexchange_prediction","tags_url":"https://repos.ecosyste.ms/api/v1/hosts/GitHub/repositories/DmitryKey%2Fkaggle_stackexchange_prediction/tags","releases_url":"https://repos.ecosyste.ms/api/v1/hosts/GitHub/repositories/DmitryKey%2Fkaggle_stackexchange_prediction/releases","manifests_url":"https://repos.ecosyste.ms/api/v1/hosts/GitHub/repositories/DmitryKey%2Fkaggle_stackexchange_prediction/manifests","owner_url":"https://repos.ecosyste.ms/api/v1/hosts/GitHub/owners/DmitryKey","download_url":"https://codeload.github.com/DmitryKey/kaggle_stackexchange_prediction/tar.gz/refs/heads/master","sbom_url":"https://repos.ecosyste.ms/api/v1/hosts/GitHub/repositories/DmitryKey%2Fkaggle_stackexchange_prediction/sbom","host":{"name":"GitHub","url":"https://github.com","kind":"github","repositories_count":266041697,"owners_count":23867944,"icon_url":"https://github.com/github.png","version":null,"created_at":"2022-05-30T11:31:42.601Z","updated_at":"2022-07-04T15:15:14.044Z","host_url":"https://repos.ecosyste.ms/api/v1/hosts/GitHub","repositories_url":"https://repos.ecosyste.ms/api/v1/hosts/GitHub/repositories","repository_names_url":"https://repos.ecosyste.ms/api/v1/hosts/GitHub/repository_names","owners_url":"https://repos.ecosyste.ms/api/v1/hosts/GitHub/owners"}},"keywords":[],"created_at":"2024-08-07T23:03:29.757Z","updated_at":"2025-07-19T23:31:52.697Z","avatar_url":"https://github.com/DmitryKey.png","language":"Python","funding_links":[],"categories":["Python"],"sub_categories":[],"readme":"CONTEST INFO\n\nThis is the source code of the submission by SkyNet team for the StackExchange contest of closed question prediction.\nThe code has originally brought us to top 25% (24th/167, 17 entries in team SkyNet), when the contest page had listed all registered teams.\nCurrent situation: we score 24-th/46 (16 entries in team SkyNet)\n\nDetails of the contest:\n\nhttp://www.kaggle.com/c/predict-closed-questions-on-stack-overflow\n\nLIBRARY\n\nWe have used scikit, pandas and numpy of Python, as well as other auxiliary helper libraries: csv, datetime, math, re and subprocess.\n\nMODELS TRAINED\n\n1. The winning algorithm was the logistic regresssion based classifier, see logit_classifier.py. \n2. Other classifiers tried didn't score any better, but are still supplied for reference and history. \nThe tried classifier algorithms: \n   Gradient Boosting\n   MNB\n   KNN\n   Perceptron (Neural Network)\n   MultiTaskElasticNet\n   ExtraTreesClassifier \n\nEVALUATION\n\nThe metric used on the contest was Multiclass Logarithmic Loss (https://www.kaggle.com/wiki/MultiClassLogLoss)\nWe implemented the metric in eval.py.\n\nWe have also ran the k-fold evaluation in order to get a sense of how the algorithm performs. This way we could do as many runs a day\nas we wanted without wasting the daily submissions (two a day). Usually our quality prediction was a little more optimistic than that of the\nactual submission, but served as a good compass in the feature selection as well as tuning the classifiers process.\n\nOTHER IDEAS (LINGUISTIC FEATURES)\n\nIn the described above classifiers we have been trying mostly the numeric features. Since a questions is written in natural language, another\nidea was to craft textual linguistic features. One such feature is a question's text perplexity calculated against the accepted and closed (rejected)\nquestions. The text perplexity can be calc'd using SRILM (see language_modelling directory). The pre-trained models of various sort are included.\n\nSome further steps might have included topical classification via extracting keywords and phrases, but wasn't tried. Feel free to try it.\n\n","project_url":"https://awesome.ecosyste.ms/api/v1/projects/github.com%2FDmitryKey%2Fkaggle_stackexchange_prediction","html_url":"https://awesome.ecosyste.ms/projects/github.com%2FDmitryKey%2Fkaggle_stackexchange_prediction","lists_url":"https://awesome.ecosyste.ms/api/v1/projects/github.com%2FDmitryKey%2Fkaggle_stackexchange_prediction/lists"}