{
  "id": 70572,
  "title": "Super high RMSE value after Submission",
  "url": "/competitions/ga-customer-revenue-prediction/discussion/70572",
  "author_name": "",
  "post_date": "2018-11-05T10:35:50.592824Z",
  "votes": null,
  "comment_count": 4,
  "views": 0,
  "content": "<p>The RMSE score for my train data was 1.59. But after I submit my result, the RMSE for test data is more than 40000+. I'm using the same model for both dataset, but getting super different result. Can anyone help me figure out what's wrong? I'm really new to this competition thing. Thanks!</p>",
  "messages": [
    {
      "id": "415570",
      "postDate": "11/05/2018 10:35:50",
      "content": "<p>The RMSE score for my train data was 1.59. But after I submit my result, the RMSE for test data is more than 40000+. I'm using the same model for both dataset, but getting super different result. Can anyone help me figure out what's wrong? I'm really new to this competition thing. Thanks!</p>",
      "rawMarkdown": "The RMSE score for my train data was 1.59. But after I submit my result, the RMSE for test data is more than 40000+. I'm using the same model for both dataset, but getting super different result. Can anyone help me figure out what's wrong? I'm really new to this competition thing. Thanks!",
      "votes": null
    },
    {
      "id": "415608",
      "postDate": "11/05/2018 11:42:26",
      "content": "<p>Try these\n1) Check if any of the predicted revenues are less than 0, reset them to zero\npred[preds &lt; 0] = 0</p>\n\n<p>2) Are you predicting the log value? Print the head of the values and see if they all are small positive values.</p>",
      "rawMarkdown": "Try these\n1) Check if any of the predicted revenues are less than 0, reset them to zero\npred[preds &lt; 0] = 0\n\n2) Are you predicting the log value? Print the head of the values and see if they all are small positive values.",
      "votes": null
    },
    {
      "id": "415836",
      "postDate": "11/05/2018 19:23:20",
      "content": "<p>Did you use expm1 to create the final output values? It might be the case that the log value is now wanted (form the notes on evaluation: 'where y hat is the natural log of the predicted revenue for a customer and y is the natural log of the actual summed revenue value plus one').</p>",
      "rawMarkdown": "Did you use expm1 to create the final output values? It might be the case that the log value is now wanted (form the notes on evaluation: 'where y hat is the natural log of the predicted revenue for a customer and y is the natural log of the actual summed revenue value plus one').",
      "votes": null
    },
    {
      "id": "415868",
      "postDate": "11/05/2018 20:36:19",
      "content": "<p>Check if you use the fullVistorId as rows or if the preds are relevant exactly to your fullVistorID?</p>",
      "rawMarkdown": "Check if you use the fullVistorId as rows or if the preds are relevant exactly to your fullVistorID?",
      "votes": null
    },
    {
      "id": "417166",
      "postDate": "11/07/2018 21:51:45",
      "content": "<p>Thanks for answering my question. I forgot to use nplog1 for the final predicted log revenue lol, pretty stupid mistake </p>",
      "rawMarkdown": "Thanks for answering my question. I forgot to use nplog1 for the final predicted log revenue lol, pretty stupid mistake",
      "votes": null
    }
  ],
  "comments": [
    {
      "id": 415608,
      "author_name": "subrahmanyamv",
      "author_url": "",
      "post_date": "11/05/2018 11:42:26",
      "content": "<p>Try these\n1) Check if any of the predicted revenues are less than 0, reset them to zero\npred[preds &lt; 0] = 0</p>\n\n<p>2) Are you predicting the log value? Print the head of the values and see if they all are small positive values.</p>",
      "votes": null,
      "replies": []
    },
    {
      "id": 415836,
      "author_name": "alltheniceai",
      "author_url": "",
      "post_date": "11/05/2018 19:23:20",
      "content": "<p>Did you use expm1 to create the final output values? It might be the case that the log value is now wanted (form the notes on evaluation: 'where y hat is the natural log of the predicted revenue for a customer and y is the natural log of the actual summed revenue value plus one').</p>",
      "votes": null,
      "replies": []
    },
    {
      "id": 415868,
      "author_name": "hurric",
      "author_url": "",
      "post_date": "11/05/2018 20:36:19",
      "content": "<p>Check if you use the fullVistorId as rows or if the preds are relevant exactly to your fullVistorID?</p>",
      "votes": null,
      "replies": []
    },
    {
      "id": 417166,
      "author_name": "kionfukuda",
      "author_url": "",
      "post_date": "11/07/2018 21:51:45",
      "content": "<p>Thanks for answering my question. I forgot to use nplog1 for the final predicted log revenue lol, pretty stupid mistake </p>",
      "votes": null,
      "replies": []
    }
  ],
  "raw_markdown_by_id": {
    "415570": "The RMSE score for my train data was 1.59. But after I submit my result, the RMSE for test data is more than 40000+. I'm using the same model for both dataset, but getting super different result. Can anyone help me figure out what's wrong? I'm really new to this competition thing. Thanks!",
    "415608": "Try these\n1) Check if any of the predicted revenues are less than 0, reset them to zero\npred[preds &lt; 0] = 0\n\n2) Are you predicting the log value? Print the head of the values and see if they all are small positive values.",
    "415836": "Did you use expm1 to create the final output values? It might be the case that the log value is now wanted (form the notes on evaluation: 'where y hat is the natural log of the predicted revenue for a customer and y is the natural log of the actual summed revenue value plus one').",
    "415868": "Check if you use the fullVistorId as rows or if the preds are relevant exactly to your fullVistorID?",
    "417166": "Thanks for answering my question. I forgot to use nplog1 for the final predicted log revenue lol, pretty stupid mistake"
  },
  "source": "meta"
}