{
  "id": 82263,
  "title": "23th -place solution (Simple CatBoost)",
  "url": "/competitions/ga-customer-revenue-prediction/discussion/82263",
  "author_name": "Xonti",
  "post_date": "2019-02-28T09:59:50.435000",
  "votes": 5,
  "comment_count": 0,
  "views": 0,
  "content": "<p>Great thanks for my first serious comtetitions. I have joined it already at the second stage so I didnt feel disappointment for sudden changes rules :)\nI used pretty standard aggregate features for my models, but i else calculated all features separately for all positive totalTransactionRevenue.\nThe absence of strong season in data allowed to use all presents data. I used each 5 month with target for 2 month pass one month. For example: x 2016/08/01 - 2017/01/01, y 2017/02/01 - 2017/04/01. For validation - just period from 2018/02/01 to 2018/08/01 with appropriate target.\nI built standart CatBoost Regression with small number iteration and it was enough to prevent overfitting. </p>",
  "messages": [
    {
      "id": 480534,
      "postDate": "2019-02-28T09:59:50.437Z",
      "content": "<p>Great thanks for my first serious comtetitions. I have joined it already at the second stage so I didnt feel disappointment for sudden changes rules :)\nI used pretty standard aggregate features for my models, but i else calculated all features separately for all positive totalTransactionRevenue.\nThe absence of strong season in data allowed to use all presents data. I used each 5 month with target for 2 month pass one month. For example: x 2016/08/01 - 2017/01/01, y 2017/02/01 - 2017/04/01. For validation - just period from 2018/02/01 to 2018/08/01 with appropriate target.\nI built standart CatBoost Regression with small number iteration and it was enough to prevent overfitting. </p>",
      "rawMarkdown": "Great thanks for my first serious comtetitions. I have joined it already at the second stage so I didnt feel disappointment for sudden changes rules :)\nI used pretty standard aggregate features for my models, but i else calculated all features separately for all positive totalTransactionRevenue.\nThe absence of strong season in data allowed to use all presents data. I used each 5 month with target for 2 month pass one month. For example: x 2016/08/01 - 2017/01/01, y 2017/02/01 - 2017/04/01. For validation - just period from 2018/02/01 to 2018/08/01 with appropriate target.\nI built standart CatBoost Regression with small number iteration and it was enough to prevent overfitting. ",
      "votes": 5
    }
  ],
  "comments": [],
  "raw_markdown_by_id": {
    "480534": "Great thanks for my first serious comtetitions. I have joined it already at the second stage so I didnt feel disappointment for sudden changes rules :)\nI used pretty standard aggregate features for my models, but i else calculated all features separately for all positive totalTransactionRevenue.\nThe absence of strong season in data allowed to use all presents data. I used each 5 month with target for 2 month pass one month. For example: x 2016/08/01 - 2017/01/01, y 2017/02/01 - 2017/04/01. For validation - just period from 2018/02/01 to 2018/08/01 with appropriate target.\nI built standart CatBoost Regression with small number iteration and it was enough to prevent overfitting. "
  },
  "complete": true,
  "source": "existing_discussion",
  "competition_id": 10038,
  "competition_slug": "ga-customer-revenue-prediction",
  "team_id": 2311778,
  "solution_url": "/competitions/ga-customer-revenue-prediction/writeups/xonti-23th-place-solution-simple-catboost"
}