{
  "id": 114474,
  "title": "submission",
  "url": "/competitions/ashrae-energy-prediction/discussion/114474",
  "author_name": "",
  "post_date": "2019-10-26T14:58:23.036324900Z",
  "votes": null,
  "comment_count": 2,
  "views": 0,
  "content": "<p>predicted meter_reading in a submission file should be normalized with log1p(meter_reading)  as values in training frame ?</p>",
  "messages": [
    {
      "id": "658801",
      "postDate": "10/26/2019 14:58:23",
      "content": "<p>predicted meter_reading in a submission file should be normalized with log1p(meter_reading)  as values in training frame ?</p>",
      "rawMarkdown": "predicted meter_reading in a submission file should be normalized with log1p(meter_reading)  as values in training frame ?",
      "votes": null
    },
    {
      "id": "658835",
      "postDate": "10/26/2019 15:52:37",
      "content": "<p>No, your submission's meter reading shoudn't be normalized. You can predict as normilized value and then transform back with np.expm1 for example.</p>",
      "rawMarkdown": "No, your submission's meter reading shoudn't be normalized. You can predict as normilized value and then transform back with np.expm1 for example.",
      "votes": null
    },
    {
      "id": "659023",
      "postDate": "10/27/2019 00:24:16",
      "content": "<p>If by 'normalized' you mean the meter_reading needs to be 'converted'  than that's correct.  If you did your work like most with log of the meter readings you need to use the exp to get back to standard units for the submission.</p>",
      "rawMarkdown": "If by 'normalized' you mean the meter_reading needs to be 'converted'  than that's correct.  If you did your work like most with log of the meter readings you need to use the exp to get back to standard units for the submission.",
      "votes": null
    }
  ],
  "comments": [
    {
      "id": 658835,
      "author_name": "ihelon",
      "author_url": "",
      "post_date": "10/26/2019 15:52:37",
      "content": "<p>No, your submission's meter reading shoudn't be normalized. You can predict as normilized value and then transform back with np.expm1 for example.</p>",
      "votes": null,
      "replies": []
    },
    {
      "id": 659023,
      "author_name": "pcjimmmy",
      "author_url": "",
      "post_date": "10/27/2019 00:24:16",
      "content": "<p>If by 'normalized' you mean the meter_reading needs to be 'converted'  than that's correct.  If you did your work like most with log of the meter readings you need to use the exp to get back to standard units for the submission.</p>",
      "votes": null,
      "replies": []
    }
  ],
  "raw_markdown_by_id": {
    "658801": "predicted meter_reading in a submission file should be normalized with log1p(meter_reading)  as values in training frame ?",
    "658835": "No, your submission's meter reading shoudn't be normalized. You can predict as normilized value and then transform back with np.expm1 for example.",
    "659023": "If by 'normalized' you mean the meter_reading needs to be 'converted'  than that's correct.  If you did your work like most with log of the meter readings you need to use the exp to get back to standard units for the submission."
  },
  "source": "meta"
}