{
  "id": 40813,
  "title": "How do you deal with the bidders that don't have any activity",
  "url": "/competitions/facebook-recruiting-iv-human-or-bot/discussion/40813",
  "author_name": "",
  "post_date": "2017-10-08T23:22:01.850684600Z",
  "votes": 1,
  "comment_count": 2,
  "views": 0,
  "content": "<p>i found some bidders that in both train and test set, who doesn't have any record on bids set. so if a bidder doesn't have any transactions, how do we generally consider this situation, do you classify them as human automatically or ? I know this is two years late, I am just a student working on this data set as practice and want to learn as much as I can. any hints and suggestion will be appreciated.</p>",
  "messages": [
    {
      "id": "229133",
      "postDate": "10/08/2017 23:22:01",
      "content": "<p>i found some bidders that in both train and test set, who doesn't have any record on bids set. so if a bidder doesn't have any transactions, how do we generally consider this situation, do you classify them as human automatically or ? I know this is two years late, I am just a student working on this data set as practice and want to learn as much as I can. any hints and suggestion will be appreciated.</p>",
      "rawMarkdown": "i found some bidders that in both train and test set, who doesn't have any record on bids set. so if a bidder doesn't have any transactions, how do we generally consider this situation, do you classify them as human automatically or ? I know this is two years late, I am just a student working on this data set as practice and want to learn as much as I can. any hints and suggestion will be appreciated.",
      "votes": null
    },
    {
      "id": "277930",
      "postDate": "02/04/2018 20:47:37",
      "content": "<p>Hello, I am a student as well. While generating features based on bidder's activity I set missing values for bidders that hadn't any activity, and finally used model based on xgboost which handles missing values internally.</p>",
      "rawMarkdown": "Hello, I am a student as well. While generating features based on bidder's activity I set missing values for bidders that hadn't any activity, and finally used model based on xgboost which handles missing values internally.",
      "votes": null
    },
    {
      "id": "432612",
      "postDate": "12/04/2018 05:15:40",
      "content": "<p>There are 70 of them in the test too! So just append NaN or something to your fused dataset and you will get better results.</p>",
      "rawMarkdown": "There are 70 of them in the test too! So just append NaN or something to your fused dataset and you will get better results.",
      "votes": null
    }
  ],
  "comments": [
    {
      "id": 277930,
      "author_name": "yleykina",
      "author_url": "",
      "post_date": "02/04/2018 20:47:37",
      "content": "<p>Hello, I am a student as well. While generating features based on bidder's activity I set missing values for bidders that hadn't any activity, and finally used model based on xgboost which handles missing values internally.</p>",
      "votes": null,
      "replies": []
    },
    {
      "id": 432612,
      "author_name": "niksp37",
      "author_url": "",
      "post_date": "12/04/2018 05:15:40",
      "content": "<p>There are 70 of them in the test too! So just append NaN or something to your fused dataset and you will get better results.</p>",
      "votes": null,
      "replies": []
    }
  ],
  "raw_markdown_by_id": {
    "229133": "i found some bidders that in both train and test set, who doesn't have any record on bids set. so if a bidder doesn't have any transactions, how do we generally consider this situation, do you classify them as human automatically or ? I know this is two years late, I am just a student working on this data set as practice and want to learn as much as I can. any hints and suggestion will be appreciated.",
    "277930": "Hello, I am a student as well. While generating features based on bidder's activity I set missing values for bidders that hadn't any activity, and finally used model based on xgboost which handles missing values internally.",
    "432612": "There are 70 of them in the test too! So just append NaN or something to your fused dataset and you will get better results."
  },
  "source": "meta"
}