{
  "id": 54825,
  "title": "Solicit advice or code review for out-of-core sklearn RF (ExtraTree)",
  "url": "/competitions/talkingdata-adtracking-fraud-detection/discussion/54825",
  "author_name": "Rene Wang",
  "post_date": "2018-04-18T08:37:09.190000",
  "votes": 0,
  "comment_count": 0,
  "views": 0,
  "content": "<p>May I have the attention of competition experts here, please.   </p>\n\n<p>I have an out-of-core scikit-learn ExtraTreeClassifier (batch mode) kernel. <br>\n<a href=\"https://www.kaggle.com/renewang/out-of-core-sklearn-extratreesclassifier\">https://www.kaggle.com/renewang/out-of-core-sklearn-extratreesclassifier</a>  </p>\n\n<p>But scikit-learn is so slow, especially using FeatureHasher. I barely to make my kernel run within one hour. No mention to do extra preprocess on features.    </p>\n\n<p>If any Kaggle guru is willing to share some ideas how to speed up with the existing code, I would be greatly grateful. </p>",
  "messages": [
    {
      "id": 316142,
      "postDate": "2018-04-18T08:37:09.190Z",
      "content": "<p>May I have the attention of competition experts here, please.   </p>\n\n<p>I have an out-of-core scikit-learn ExtraTreeClassifier (batch mode) kernel. <br>\n<a href=\"https://www.kaggle.com/renewang/out-of-core-sklearn-extratreesclassifier\">https://www.kaggle.com/renewang/out-of-core-sklearn-extratreesclassifier</a>  </p>\n\n<p>But scikit-learn is so slow, especially using FeatureHasher. I barely to make my kernel run within one hour. No mention to do extra preprocess on features.    </p>\n\n<p>If any Kaggle guru is willing to share some ideas how to speed up with the existing code, I would be greatly grateful. </p>",
      "rawMarkdown": "May I have the attention of competition experts here, please.   \n\nI have an out-of-core scikit-learn ExtraTreeClassifier (batch mode) kernel.   \nhttps://www.kaggle.com/renewang/out-of-core-sklearn-extratreesclassifier  \n\nBut scikit-learn is so slow, especially using FeatureHasher. I barely to make my kernel run within one hour. No mention to do extra preprocess on features.    \n\nIf any Kaggle guru is willing to share some ideas how to speed up with the existing code, I would be greatly grateful. "
    }
  ],
  "comments": [],
  "raw_markdown_by_id": {
    "316142": "May I have the attention of competition experts here, please.   \n\nI have an out-of-core scikit-learn ExtraTreeClassifier (batch mode) kernel.   \nhttps://www.kaggle.com/renewang/out-of-core-sklearn-extratreesclassifier  \n\nBut scikit-learn is so slow, especially using FeatureHasher. I barely to make my kernel run within one hour. No mention to do extra preprocess on features.    \n\nIf any Kaggle guru is willing to share some ideas how to speed up with the existing code, I would be greatly grateful. "
  }
}