{
  "id": 82333,
  "title": "Rebalancing Anyone",
  "url": "/competitions/elo-merchant-category-recommendation/discussion/82333",
  "author_name": "",
  "post_date": "2019-02-28T21:24:55.773267100Z",
  "votes": null,
  "comment_count": 1,
  "views": 0,
  "content": "<p>Hi All, \nDid anyone try to rebalance the training set, considering the outliers, with undersampling, oversampling or  SMOTE or any other technique? I did and results did not improve at all. </p>",
  "messages": [
    {
      "id": "480918",
      "postDate": "02/28/2019 21:24:55",
      "content": "<p>Hi All, \nDid anyone try to rebalance the training set, considering the outliers, with undersampling, oversampling or  SMOTE or any other technique? I did and results did not improve at all. </p>",
      "rawMarkdown": "Hi All, \nDid anyone try to rebalance the training set, considering the outliers, with undersampling, oversampling or  SMOTE or any other technique? I did and results did not improve at all.",
      "votes": null
    },
    {
      "id": "480920",
      "postDate": "02/28/2019 21:33:59",
      "content": "<p>I used undersampling using 80 sets of all outliers + same number of random nonoutliers for LGBM classifier, took the mean of all of their test set predictions.\nthis helped a bit in stacking with few models, but lost its value when stacking was used with many more models.</p>",
      "rawMarkdown": "I used undersampling using 80 sets of all outliers + same number of random nonoutliers for LGBM classifier, took the mean of all of their test set predictions.\nthis helped a bit in stacking with few models, but lost its value when stacking was used with many more models.",
      "votes": null
    }
  ],
  "comments": [
    {
      "id": 480920,
      "author_name": "tpthegreat",
      "author_url": "",
      "post_date": "02/28/2019 21:33:59",
      "content": "<p>I used undersampling using 80 sets of all outliers + same number of random nonoutliers for LGBM classifier, took the mean of all of their test set predictions.\nthis helped a bit in stacking with few models, but lost its value when stacking was used with many more models.</p>",
      "votes": null,
      "replies": []
    }
  ],
  "raw_markdown_by_id": {
    "480918": "Hi All, \nDid anyone try to rebalance the training set, considering the outliers, with undersampling, oversampling or  SMOTE or any other technique? I did and results did not improve at all.",
    "480920": "I used undersampling using 80 sets of all outliers + same number of random nonoutliers for LGBM classifier, took the mean of all of their test set predictions.\nthis helped a bit in stacking with few models, but lost its value when stacking was used with many more models."
  },
  "source": "meta"
}