{
  "id": 76001,
  "title": "Why month_diff feature so important",
  "url": "/competitions/elo-merchant-category-recommendation/discussion/76001",
  "author_name": "",
  "post_date": "2018-12-28T11:20:46.250836Z",
  "votes": 5,
  "comment_count": 1,
  "views": 0,
  "content": "<p>Many kernels have created the month_diff feature, which is the difference between today and each purchase reference date. I used this feature to get a improve of RMSE about 0.002. Dose any one have idea about why this feature is so significant </p>",
  "messages": [
    {
      "id": "446605",
      "postDate": "12/28/2018 11:20:46",
      "content": "<p>Many kernels have created the month_diff feature, which is the difference between today and each purchase reference date. I used this feature to get a improve of RMSE about 0.002. Dose any one have idea about why this feature is so significant </p>",
      "rawMarkdown": "Many kernels have created the month_diff feature, which is the difference between today and each purchase reference date. I used this feature to get a improve of RMSE about 0.002. Dose any one have idea about why this feature is so significant",
      "votes": null
    },
    {
      "id": "446750",
      "postDate": "12/28/2018 16:25:04",
      "content": "<p>Hi! Luuunh, I believe that the Loyalty Score calculation in some way has more weight towards the most recent purchases or closer to some date, so if you expose this feature to the model it has an easy time finding correlations, just my theory </p>",
      "rawMarkdown": "Hi! Luuunh, I believe that the Loyalty Score calculation in some way has more weight towards the most recent purchases or closer to some date, so if you expose this feature to the model it has an easy time finding correlations, just my theory",
      "votes": null
    }
  ],
  "comments": [
    {
      "id": 446750,
      "author_name": "cv13j0",
      "author_url": "",
      "post_date": "12/28/2018 16:25:04",
      "content": "<p>Hi! Luuunh, I believe that the Loyalty Score calculation in some way has more weight towards the most recent purchases or closer to some date, so if you expose this feature to the model it has an easy time finding correlations, just my theory </p>",
      "votes": null,
      "replies": []
    }
  ],
  "raw_markdown_by_id": {
    "446605": "Many kernels have created the month_diff feature, which is the difference between today and each purchase reference date. I used this feature to get a improve of RMSE about 0.002. Dose any one have idea about why this feature is so significant",
    "446750": "Hi! Luuunh, I believe that the Loyalty Score calculation in some way has more weight towards the most recent purchases or closer to some date, so if you expose this feature to the model it has an easy time finding correlations, just my theory"
  },
  "source": "meta"
}