{
  "id": 113926,
  "title": "How do you think predicting zeroes in target variable?",
  "url": "/competitions/ashrae-energy-prediction/discussion/113926",
  "author_name": "",
  "post_date": "2019-10-23T04:06:10.648858200Z",
  "votes": 2,
  "comment_count": 5,
  "views": 0,
  "content": "<p>When I just faced zeroes in target variable, I just thought the meter_reader(?) went wrong. So I just thought I could predict target variable's 'zeroes' with ML. Is this a stupid idea?</p>",
  "messages": [
    {
      "id": "655444",
      "postDate": "10/23/2019 04:06:10",
      "content": "<p>When I just faced zeroes in target variable, I just thought the meter_reader(?) went wrong. So I just thought I could predict target variable's 'zeroes' with ML. Is this a stupid idea?</p>",
      "rawMarkdown": "When I just faced zeroes in target variable, I just thought the meter_reader(?) went wrong. So I just thought I could predict target variable's 'zeroes' with ML. Is this a stupid idea?",
      "votes": null
    },
    {
      "id": "655449",
      "postDate": "10/23/2019 04:17:23",
      "content": "<ul>\n<li>I guess the reason why there are some zeros in target variables is just that these building had not set the sensor at the time. I think it is because these zeros occurred at the beginning of observations. I have to check whether counter-examples exist or not.</li>\n</ul>",
      "rawMarkdown": "I guess the reason why there are some zeros in target variables is just that these building had not set the sensor at the time. I think it is because these zeros occurred at the beginning of observations. I have to check whether counter-examples exist or not.",
      "votes": null
    },
    {
      "id": "655761",
      "postDate": "10/23/2019 13:30:33",
      "content": "<p>For building id = 7 and meter = 1 there are sudden zeros in between the year and also reading started in 3rd month of the year. \n<img src=\"https://www.googleapis.com/download/storage/v1/b/kaggle-user-content/o/inbox%2F1148850%2Ff02507348b80b557acb472d1e9b2384e%2Ffor_id_7.png?generation=1571837399461024&amp;alt=media\" alt=\"\"></p>",
      "rawMarkdown": "For building id = 7 and meter = 1 there are sudden zeros in between the year and also reading started in 3rd month of the year. \n![](https://www.googleapis.com/download/storage/v1/b/kaggle-user-content/o/inbox%2F1148850%2Ff02507348b80b557acb472d1e9b2384e%2Ffor_id_7.png?generation=1571837399461024&amp;alt=media)",
      "votes": null
    },
    {
      "id": "655957",
      "postDate": "10/23/2019 18:01:43",
      "content": "<p>Thanks for the information, I think there might be great things to do here like filling gaps or erasing them for example 👀 😏 </p>",
      "rawMarkdown": "Thanks for the information, I think there might be great things to do here like filling gaps or erasing them for example 👀 😏",
      "votes": null
    },
    {
      "id": "656018",
      "postDate": "10/23/2019 20:05:17",
      "content": "<p><a href=\"/suchith0312\">@suchith0312</a>  Great EDA. Can you share you code for this?</p>",
      "rawMarkdown": "suchith0312  Great EDA. Can you share you code for this?",
      "votes": null
    },
    {
      "id": "656326",
      "postDate": "10/24/2019 06:41:08",
      "content": "<p><code>\nx = train.loc[(train['building_id'] == 7) &amp; (train['meter'] == 0)]\nplt.figure(figsize=(15,5))\nplt.xticks(rotation='90')\nplt.plot(x['timestamp'],np.log1p(x['meter_reading']),label='meter = 0')\nx = train.loc[(train['building_id'] == 7) &amp; (train['meter'] == 1)]\nplt.plot(x['timestamp'],np.log1p(x['meter_reading']),label='meter = 1')\nplt.title(\"building id = 7\")\nplt.legend()\nplt.show()\n</code></p>",
      "rawMarkdown": "```\nx = train.loc[(train['building_id'] == 7) &amp; (train['meter'] == 0)]\nplt.figure(figsize=(15,5))\nplt.xticks(rotation='90')\nplt.plot(x['timestamp'],np.log1p(x['meter_reading']),label='meter = 0')\nx = train.loc[(train['building_id'] == 7) &amp; (train['meter'] == 1)]\nplt.plot(x['timestamp'],np.log1p(x['meter_reading']),label='meter = 1')\nplt.title(\"building id = 7\")\nplt.legend()\nplt.show()\n```",
      "votes": null
    }
  ],
  "comments": [
    {
      "id": 655449,
      "author_name": "econjt",
      "author_url": "",
      "post_date": "10/23/2019 04:17:23",
      "content": "<ul>\n<li>I guess the reason why there are some zeros in target variables is just that these building had not set the sensor at the time. I think it is because these zeros occurred at the beginning of observations. I have to check whether counter-examples exist or not.</li>\n</ul>",
      "votes": null,
      "replies": [
        {
          "id": 655761,
          "author_name": "suchith0312",
          "author_url": "",
          "post_date": "10/23/2019 13:30:33",
          "content": "<p>For building id = 7 and meter = 1 there are sudden zeros in between the year and also reading started in 3rd month of the year. \n<img src=\"https://www.googleapis.com/download/storage/v1/b/kaggle-user-content/o/inbox%2F1148850%2Ff02507348b80b557acb472d1e9b2384e%2Ffor_id_7.png?generation=1571837399461024&amp;alt=media\" alt=\"\"></p>",
          "votes": null,
          "replies": []
        },
        {
          "id": 655957,
          "author_name": "drexpz",
          "author_url": "",
          "post_date": "10/23/2019 18:01:43",
          "content": "<p>Thanks for the information, I think there might be great things to do here like filling gaps or erasing them for example 👀 😏 </p>",
          "votes": null,
          "replies": []
        },
        {
          "id": 656018,
          "author_name": "themonologue",
          "author_url": "",
          "post_date": "10/23/2019 20:05:17",
          "content": "<p><a href=\"/suchith0312\">@suchith0312</a>  Great EDA. Can you share you code for this?</p>",
          "votes": null,
          "replies": []
        },
        {
          "id": 656326,
          "author_name": "suchith0312",
          "author_url": "",
          "post_date": "10/24/2019 06:41:08",
          "content": "<p><code>\nx = train.loc[(train['building_id'] == 7) &amp; (train['meter'] == 0)]\nplt.figure(figsize=(15,5))\nplt.xticks(rotation='90')\nplt.plot(x['timestamp'],np.log1p(x['meter_reading']),label='meter = 0')\nx = train.loc[(train['building_id'] == 7) &amp; (train['meter'] == 1)]\nplt.plot(x['timestamp'],np.log1p(x['meter_reading']),label='meter = 1')\nplt.title(\"building id = 7\")\nplt.legend()\nplt.show()\n</code></p>",
          "votes": null,
          "replies": []
        }
      ]
    }
  ],
  "raw_markdown_by_id": {
    "655444": "When I just faced zeroes in target variable, I just thought the meter_reader(?) went wrong. So I just thought I could predict target variable's 'zeroes' with ML. Is this a stupid idea?",
    "655449": "I guess the reason why there are some zeros in target variables is just that these building had not set the sensor at the time. I think it is because these zeros occurred at the beginning of observations. I have to check whether counter-examples exist or not.",
    "655761": "For building id = 7 and meter = 1 there are sudden zeros in between the year and also reading started in 3rd month of the year. \n![](https://www.googleapis.com/download/storage/v1/b/kaggle-user-content/o/inbox%2F1148850%2Ff02507348b80b557acb472d1e9b2384e%2Ffor_id_7.png?generation=1571837399461024&amp;alt=media)",
    "655957": "Thanks for the information, I think there might be great things to do here like filling gaps or erasing them for example 👀 😏",
    "656018": "suchith0312  Great EDA. Can you share you code for this?",
    "656326": "```\nx = train.loc[(train['building_id'] == 7) &amp; (train['meter'] == 0)]\nplt.figure(figsize=(15,5))\nplt.xticks(rotation='90')\nplt.plot(x['timestamp'],np.log1p(x['meter_reading']),label='meter = 0')\nx = train.loc[(train['building_id'] == 7) &amp; (train['meter'] == 1)]\nplt.plot(x['timestamp'],np.log1p(x['meter_reading']),label='meter = 1')\nplt.title(\"building id = 7\")\nplt.legend()\nplt.show()\n```"
  },
  "source": "meta"
}