{
  "id": 59041,
  "title": "could you share your lgbm paramter settings? ",
  "url": "/competitions/avito-demand-prediction/discussion/59041",
  "author_name": "",
  "post_date": "2018-06-17T15:17:00.911420900Z",
  "votes": null,
  "comment_count": 3,
  "views": 0,
  "content": "<p>hou much rounds, leafs, depth and learning rate etc.</p>",
  "messages": [
    {
      "id": "344304",
      "postDate": "06/17/2018 15:17:00",
      "content": "<p>hou much rounds, leafs, depth and learning rate etc.</p>",
      "rawMarkdown": "hou much rounds, leafs, depth and learning rate etc.",
      "votes": null
    },
    {
      "id": "344352",
      "postDate": "06/17/2018 18:21:18",
      "content": "<p><a href=\"http://scikit-learn.org/stable/modules/generated/sklearn.model_selection.GridSearchCV.html\">http://scikit-learn.org/stable/modules/generated/sklearn.model_selection.GridSearchCV.html</a> may be of some help. </p>",
      "rawMarkdown": "http://scikit-learn.org/stable/modules/generated/sklearn.model_selection.GridSearchCV.html may be of some help.",
      "votes": null
    },
    {
      "id": "344564",
      "postDate": "06/18/2018 08:21:45",
      "content": "<p>I think it depends more on your data... You can use Grid Search with a more coarse values and then fine tune with another round... Others parameters may not be useful for your data... For a general idea you can see the public kernels that used LGB...</p>",
      "rawMarkdown": "I think it depends more on your data... You can use Grid Search with a more coarse values and then fine tune with another round... Others parameters may not be useful for your data... For a general idea you can see the public kernels that used LGB...",
      "votes": null
    },
    {
      "id": "344573",
      "postDate": "06/18/2018 08:43:42",
      "content": "<p><strong>LightGBM parameters found by Bayesian optimization</strong></p>\n\n<pre><code>    clf = LGBMClassifier(\n        nthread=4,\n        n_estimators=10000,\n        learning_rate=0.02,\n        num_leaves=34,\n        colsample_bytree=0.9497036,\n        subsample=0.8715623,\n        max_depth=8,\n        reg_alpha=0.041545473,\n        reg_lambda=0.0735294,\n        min_split_gain=0.0222415,\n        min_child_weight=39.3259775,\n        silent=-1,\n        verbose=-1, )\n</code></pre>",
      "rawMarkdown": "**LightGBM parameters found by Bayesian optimization**\n\n\n        clf = LGBMClassifier(\n            nthread=4,\n            n_estimators=10000,\n            learning_rate=0.02,\n            num_leaves=34,\n            colsample_bytree=0.9497036,\n            subsample=0.8715623,\n            max_depth=8,\n            reg_alpha=0.041545473,\n            reg_lambda=0.0735294,\n            min_split_gain=0.0222415,\n            min_child_weight=39.3259775,\n            silent=-1,\n            verbose=-1, )",
      "votes": null
    }
  ],
  "comments": [
    {
      "id": 344352,
      "author_name": "abrial",
      "author_url": "",
      "post_date": "06/17/2018 18:21:18",
      "content": "<p><a href=\"http://scikit-learn.org/stable/modules/generated/sklearn.model_selection.GridSearchCV.html\">http://scikit-learn.org/stable/modules/generated/sklearn.model_selection.GridSearchCV.html</a> may be of some help. </p>",
      "votes": null,
      "replies": []
    },
    {
      "id": 344564,
      "author_name": "samratp",
      "author_url": "",
      "post_date": "06/18/2018 08:21:45",
      "content": "<p>I think it depends more on your data... You can use Grid Search with a more coarse values and then fine tune with another round... Others parameters may not be useful for your data... For a general idea you can see the public kernels that used LGB...</p>",
      "votes": null,
      "replies": []
    },
    {
      "id": 344573,
      "author_name": "ashishpatel26",
      "author_url": "",
      "post_date": "06/18/2018 08:43:42",
      "content": "<p><strong>LightGBM parameters found by Bayesian optimization</strong></p>\n\n<pre><code>    clf = LGBMClassifier(\n        nthread=4,\n        n_estimators=10000,\n        learning_rate=0.02,\n        num_leaves=34,\n        colsample_bytree=0.9497036,\n        subsample=0.8715623,\n        max_depth=8,\n        reg_alpha=0.041545473,\n        reg_lambda=0.0735294,\n        min_split_gain=0.0222415,\n        min_child_weight=39.3259775,\n        silent=-1,\n        verbose=-1, )\n</code></pre>",
      "votes": null,
      "replies": []
    }
  ],
  "raw_markdown_by_id": {
    "344304": "hou much rounds, leafs, depth and learning rate etc.",
    "344352": "http://scikit-learn.org/stable/modules/generated/sklearn.model_selection.GridSearchCV.html may be of some help.",
    "344564": "I think it depends more on your data... You can use Grid Search with a more coarse values and then fine tune with another round... Others parameters may not be useful for your data... For a general idea you can see the public kernels that used LGB...",
    "344573": "**LightGBM parameters found by Bayesian optimization**\n\n\n        clf = LGBMClassifier(\n            nthread=4,\n            n_estimators=10000,\n            learning_rate=0.02,\n            num_leaves=34,\n            colsample_bytree=0.9497036,\n            subsample=0.8715623,\n            max_depth=8,\n            reg_alpha=0.041545473,\n            reg_lambda=0.0735294,\n            min_split_gain=0.0222415,\n            min_child_weight=39.3259775,\n            silent=-1,\n            verbose=-1, )"
  },
  "source": "meta"
}