{
  "id": 107364,
  "title": "How to ensemble?",
  "url": "/competitions/recursion-cellular-image-classification/discussion/107364",
  "author_name": "",
  "post_date": "2019-09-03T23:34:41.428916900Z",
  "votes": 4,
  "comment_count": 6,
  "views": 0,
  "content": "<p>Hi guys, do you mind sharing how to ensemble models?</p>\n\n<p>I wonder is it possible to take the prediction file across all models and just sum them together. In other words, the <code>(19897, 1108)</code> prediction file from a model added to the <code>(19897, 1108)</code> prediction file of another model and then take the <code>argmax</code> for each row. Is this correct?</p>\n\n<p>Thank you.</p>",
  "messages": [
    {
      "id": "617241",
      "postDate": "09/03/2019 23:34:41",
      "content": "<p>Hi guys, do you mind sharing how to ensemble models?</p>\n\n<p>I wonder is it possible to take the prediction file across all models and just sum them together. In other words, the <code>(19897, 1108)</code> prediction file from a model added to the <code>(19897, 1108)</code> prediction file of another model and then take the <code>argmax</code> for each row. Is this correct?</p>\n\n<p>Thank you.</p>",
      "rawMarkdown": "Hi guys, do you mind sharing how to ensemble models?\n\nI wonder is it possible to take the prediction file across all models and just sum them together. In other words, the `(19897, 1108)` prediction file from a model added to the `(19897, 1108)` prediction file of another model and then take the `argmax` for each row. Is this correct?\n\nThank you.",
      "votes": null
    },
    {
      "id": "617418",
      "postDate": "09/04/2019 05:33:35",
      "content": "<p>That's the common way</p>",
      "rawMarkdown": "That's the common way",
      "votes": null
    },
    {
      "id": "617424",
      "postDate": "09/04/2019 05:44:29",
      "content": "<p>This is a generic answer, but if you have never read this one I recommend to take a look: <a href=\"https://mlwave.com/kaggle-ensembling-guide/\">Kaggle Ensembling Guide</a>.</p>",
      "rawMarkdown": "This is a generic answer, but if you have never read this one I recommend to take a look: [Kaggle Ensembling Guide](https://mlwave.com/kaggle-ensembling-guide/).",
      "votes": null
    },
    {
      "id": "617537",
      "postDate": "09/04/2019 08:19:06",
      "content": "<p><a href=\"https://machinelearningmastery.com/stacking-ensemble-for-deep-learning-neural-networks/\">https://machinelearningmastery.com/stacking-ensemble-for-deep-learning-neural-networks/</a></p>",
      "rawMarkdown": "https://machinelearningmastery.com/stacking-ensemble-for-deep-learning-neural-networks/",
      "votes": null
    },
    {
      "id": "617538",
      "postDate": "09/04/2019 08:21:19",
      "content": "<p>very good article,i went through it few days later then tried mlwave by cloning codes from my repository but i couldn't run it in my windows 10 computer,the problem was with argv i think,i went through stackoverflow posts but nothing helped,have you ever used it? can you please tell me how to do it in kaggle?</p>",
      "rawMarkdown": "very good article,i went through it few days later then tried mlwave by cloning codes from my repository but i couldn't run it in my windows 10 computer,the problem was with argv i think,i went through stackoverflow posts but nothing helped,have you ever used it? can you please tell me how to do it in kaggle?",
      "votes": null
    },
    {
      "id": "617610",
      "postDate": "09/04/2019 10:07:09",
      "content": "<p>I have never tried to use his code, just read the article itself.</p>",
      "rawMarkdown": "I have never tried to use his code, just read the article itself.",
      "votes": null
    },
    {
      "id": "617793",
      "postDate": "09/04/2019 13:41:23",
      "content": "<p>Thats a legit way to do it. Just make sure the predictions are of a comparable scale, ie, not such that one set prediction is between 0 and 1 while another may be -inf to inf.</p>",
      "rawMarkdown": "Thats a legit way to do it. Just make sure the predictions are of a comparable scale, ie, not such that one set prediction is between 0 and 1 while another may be -inf to inf.",
      "votes": null
    }
  ],
  "comments": [
    {
      "id": 617418,
      "author_name": "valanm",
      "author_url": "",
      "post_date": "09/04/2019 05:33:35",
      "content": "<p>That's the common way</p>",
      "votes": null,
      "replies": []
    },
    {
      "id": 617424,
      "author_name": "zaharch",
      "author_url": "",
      "post_date": "09/04/2019 05:44:29",
      "content": "<p>This is a generic answer, but if you have never read this one I recommend to take a look: <a href=\"https://mlwave.com/kaggle-ensembling-guide/\">Kaggle Ensembling Guide</a>.</p>",
      "votes": null,
      "replies": [
        {
          "id": 617538,
          "author_name": "mobassir",
          "author_url": "",
          "post_date": "09/04/2019 08:21:19",
          "content": "<p>very good article,i went through it few days later then tried mlwave by cloning codes from my repository but i couldn't run it in my windows 10 computer,the problem was with argv i think,i went through stackoverflow posts but nothing helped,have you ever used it? can you please tell me how to do it in kaggle?</p>",
          "votes": null,
          "replies": []
        },
        {
          "id": 617610,
          "author_name": "zaharch",
          "author_url": "",
          "post_date": "09/04/2019 10:07:09",
          "content": "<p>I have never tried to use his code, just read the article itself.</p>",
          "votes": null,
          "replies": []
        }
      ]
    },
    {
      "id": 617537,
      "author_name": "mobassir",
      "author_url": "",
      "post_date": "09/04/2019 08:19:06",
      "content": "<p><a href=\"https://machinelearningmastery.com/stacking-ensemble-for-deep-learning-neural-networks/\">https://machinelearningmastery.com/stacking-ensemble-for-deep-learning-neural-networks/</a></p>",
      "votes": null,
      "replies": []
    },
    {
      "id": 617793,
      "author_name": "interneuron",
      "author_url": "",
      "post_date": "09/04/2019 13:41:23",
      "content": "<p>Thats a legit way to do it. Just make sure the predictions are of a comparable scale, ie, not such that one set prediction is between 0 and 1 while another may be -inf to inf.</p>",
      "votes": null,
      "replies": []
    }
  ],
  "raw_markdown_by_id": {
    "617241": "Hi guys, do you mind sharing how to ensemble models?\n\nI wonder is it possible to take the prediction file across all models and just sum them together. In other words, the `(19897, 1108)` prediction file from a model added to the `(19897, 1108)` prediction file of another model and then take the `argmax` for each row. Is this correct?\n\nThank you.",
    "617418": "That's the common way",
    "617424": "This is a generic answer, but if you have never read this one I recommend to take a look: [Kaggle Ensembling Guide](https://mlwave.com/kaggle-ensembling-guide/).",
    "617537": "https://machinelearningmastery.com/stacking-ensemble-for-deep-learning-neural-networks/",
    "617538": "very good article,i went through it few days later then tried mlwave by cloning codes from my repository but i couldn't run it in my windows 10 computer,the problem was with argv i think,i went through stackoverflow posts but nothing helped,have you ever used it? can you please tell me how to do it in kaggle?",
    "617610": "I have never tried to use his code, just read the article itself.",
    "617793": "Thats a legit way to do it. Just make sure the predictions are of a comparable scale, ie, not such that one set prediction is between 0 and 1 while another may be -inf to inf."
  },
  "source": "meta"
}