{
  "id": 94240,
  "title": "Problems with submission - what's wrong?",
  "url": "/competitions/champs-scalar-coupling/discussion/94240",
  "author_name": "",
  "post_date": "2019-06-03T12:06:56.320265600Z",
  "votes": 1,
  "comment_count": 2,
  "views": 0,
  "content": "<p>Hi Guys, I'm playing on this competition and trying to do a good submission, but when I train my model, I got 0.4~~ score using the following metric:</p>\n\n<p>def group_mean_log_mae(y_true, y_pred, types, floor=1e-9):\n    maes = (y_true-y_pred).abs().groupby(types).mean()\n    return np.log(maes.map(lambda x: max(x, floor))).mean()</p>\n\n<p>But when I run the prediction with the test set I got a bizarre score.</p>\n\n<p>Could I do something before predict the molecular properties? What I'm doing wrong?</p>\n\n<p>Thank you very much</p>",
  "messages": [
    {
      "id": "542066",
      "postDate": "06/03/2019 12:06:56",
      "content": "<p>Hi Guys, I'm playing on this competition and trying to do a good submission, but when I train my model, I got 0.4~~ score using the following metric:</p>\n\n<p>def group_mean_log_mae(y_true, y_pred, types, floor=1e-9):\n    maes = (y_true-y_pred).abs().groupby(types).mean()\n    return np.log(maes.map(lambda x: max(x, floor))).mean()</p>\n\n<p>But when I run the prediction with the test set I got a bizarre score.</p>\n\n<p>Could I do something before predict the molecular properties? What I'm doing wrong?</p>\n\n<p>Thank you very much</p>",
      "rawMarkdown": "Hi Guys, I'm playing on this competition and trying to do a good submission, but when I train my model, I got 0.4~~ score using the following metric:\n\ndef group_mean_log_mae(y_true, y_pred, types, floor=1e-9):\n    maes = (y_true-y_pred).abs().groupby(types).mean()\n    return np.log(maes.map(lambda x: max(x, floor))).mean()\n\nBut when I run the prediction with the test set I got a bizarre score.\n\nCould I do something before predict the molecular properties? What I'm doing wrong?\n\nThank you very much",
      "votes": null
    },
    {
      "id": "542429",
      "postDate": "06/03/2019 22:53:14",
      "content": "<p>Maybe a feature in the test set in messed up, but the feature is fine in the training set.  This would cause oof predictions to be okay, but the submission will be off.</p>",
      "rawMarkdown": "Maybe a feature in the test set in messed up, but the feature is fine in the training set.  This would cause oof predictions to be okay, but the submission will be off.",
      "votes": null
    },
    {
      "id": "542460",
      "postDate": "06/04/2019 00:11:15",
      "content": "<p>Hi Kyle, it's exactly what I discovered LOL\nI had forgotten to transform the test df; Thank you very much</p>",
      "rawMarkdown": "Hi Kyle, it's exactly what I discovered LOL\nI had forgotten to transform the test df; Thank you very much",
      "votes": null
    }
  ],
  "comments": [
    {
      "id": 542429,
      "author_name": "petersk20",
      "author_url": "",
      "post_date": "06/03/2019 22:53:14",
      "content": "<p>Maybe a feature in the test set in messed up, but the feature is fine in the training set.  This would cause oof predictions to be okay, but the submission will be off.</p>",
      "votes": null,
      "replies": [
        {
          "id": 542460,
          "author_name": "kabure",
          "author_url": "",
          "post_date": "06/04/2019 00:11:15",
          "content": "<p>Hi Kyle, it's exactly what I discovered LOL\nI had forgotten to transform the test df; Thank you very much</p>",
          "votes": null,
          "replies": []
        }
      ]
    }
  ],
  "raw_markdown_by_id": {
    "542066": "Hi Guys, I'm playing on this competition and trying to do a good submission, but when I train my model, I got 0.4~~ score using the following metric:\n\ndef group_mean_log_mae(y_true, y_pred, types, floor=1e-9):\n    maes = (y_true-y_pred).abs().groupby(types).mean()\n    return np.log(maes.map(lambda x: max(x, floor))).mean()\n\nBut when I run the prediction with the test set I got a bizarre score.\n\nCould I do something before predict the molecular properties? What I'm doing wrong?\n\nThank you very much",
    "542429": "Maybe a feature in the test set in messed up, but the feature is fine in the training set.  This would cause oof predictions to be okay, but the submission will be off.",
    "542460": "Hi Kyle, it's exactly what I discovered LOL\nI had forgotten to transform the test df; Thank you very much"
  },
  "source": "meta"
}