{
  "id": 100267,
  "title": "Model Ensembles?",
  "url": "/competitions/aptos2019-blindness-detection/discussion/100267",
  "author_name": "",
  "post_date": "2019-07-17T14:26:15.514256Z",
  "votes": null,
  "comment_count": 2,
  "views": 0,
  "content": "<p>Has anyone tried a Model Ensemble so far, and choosing the weight of every model with Ridge Regression? \nI saw this method before here: <a href=\"http://blog.kaggle.com/2017/10/17/planet-understanding-the-amazon-from-space-1st-place-winners-interview/\">http://blog.kaggle.com/2017/10/17/planet-understanding-the-amazon-from-space-1st-place-winners-interview/</a>\nHowever, it seems like a little of a computational overkill to me.</p>\n\n<p>Any thoughts on this?</p>",
  "messages": [
    {
      "id": "578260",
      "postDate": "07/17/2019 14:26:15",
      "content": "<p>Has anyone tried a Model Ensemble so far, and choosing the weight of every model with Ridge Regression? \nI saw this method before here: <a href=\"http://blog.kaggle.com/2017/10/17/planet-understanding-the-amazon-from-space-1st-place-winners-interview/\">http://blog.kaggle.com/2017/10/17/planet-understanding-the-amazon-from-space-1st-place-winners-interview/</a>\nHowever, it seems like a little of a computational overkill to me.</p>\n\n<p>Any thoughts on this?</p>",
      "rawMarkdown": "Has anyone tried a Model Ensemble so far, and choosing the weight of every model with Ridge Regression? \nI saw this method before here: http://blog.kaggle.com/2017/10/17/planet-understanding-the-amazon-from-space-1st-place-winners-interview/\nHowever, it seems like a little of a computational overkill to me.\n\nAny thoughts on this?",
      "votes": null
    },
    {
      "id": "578654",
      "postDate": "07/18/2019 01:45:37",
      "content": "<p>Being a kernels-only competition, it will be difficult to have a lot of models for ensemble, unless they have low inference times.</p>",
      "rawMarkdown": "Being a kernels-only competition, it will be difficult to have a lot of models for ensemble, unless they have low inference times.",
      "votes": null
    },
    {
      "id": "578752",
      "postDate": "07/18/2019 05:25:22",
      "content": "<p>fair point, thanks for your answer!</p>",
      "rawMarkdown": "fair point, thanks for your answer!",
      "votes": null
    }
  ],
  "comments": [
    {
      "id": 578654,
      "author_name": "tanlikesmath",
      "author_url": "",
      "post_date": "07/18/2019 01:45:37",
      "content": "<p>Being a kernels-only competition, it will be difficult to have a lot of models for ensemble, unless they have low inference times.</p>",
      "votes": null,
      "replies": [
        {
          "id": 578752,
          "author_name": "fanconic",
          "author_url": "",
          "post_date": "07/18/2019 05:25:22",
          "content": "<p>fair point, thanks for your answer!</p>",
          "votes": null,
          "replies": []
        }
      ]
    }
  ],
  "raw_markdown_by_id": {
    "578260": "Has anyone tried a Model Ensemble so far, and choosing the weight of every model with Ridge Regression? \nI saw this method before here: http://blog.kaggle.com/2017/10/17/planet-understanding-the-amazon-from-space-1st-place-winners-interview/\nHowever, it seems like a little of a computational overkill to me.\n\nAny thoughts on this?",
    "578654": "Being a kernels-only competition, it will be difficult to have a lot of models for ensemble, unless they have low inference times.",
    "578752": "fair point, thanks for your answer!"
  },
  "source": "meta"
}