{
  "id": 81496,
  "title": "110th Place Solution",
  "url": "/competitions/ga-customer-revenue-prediction/writeups/tak-110th-place-solution",
  "author_name": "",
  "post_date": "2019-02-22T07:01:45.123Z",
  "votes": 8,
  "comment_count": 2,
  "views": 0,
  "content": "<p>I briefly write down my solution, which got medal-border score. </p>\n\n<p>So, this is not a complex approach.</p>\n\n<blockquote>\n  <ul>\n  <li><strong>Feature Engineering</strong>: based on <a href=\"https://www.kaggle.com/yoshoku/gacrp-v2-starter-kit\">this kernel</a></li>\n  <li><strong>Modeling...Classification =&gt; Regression</strong>: As discussed <a href=\"https://www.kaggle.com/c/ga-customer-revenue-prediction/discussion/71427\">here</a>, after classifying whether the re-visited purchase will occur or not, I made a regression model which predicts how much the amount of the purchase is.</li>\n  <li><strong>Task Timeframe</strong>: In this task, we have to predict the amount of the purchase from 2018/12/01 to 2019/1/31 per user who has visited between 2018/5/1 and 2018/10/15 with the data whose timeframe is from 2016/4/1 to 2018/4/30 and from 2018/5/1 to 2018/10/15.</li>\n  <li><strong>Validation Strategy</strong>: To consider the problem simply, aiming to ignore the seasonality, I decided to take the following validation strategy: <br>\n  <ul><li>The task is to predict whether re-visited purchase will occur or not between 2018/12/1 and 2019/1/31 by using data whose timeframe is from 2018/5/1 to 2018/10/15.  </li>\n  <li>For validation, I predicted the probability of the occurrance of re-visit between 2017/12/1 and 2018/1/31 by using data whose timeframe is from 2017/5/1 to 2017/10/15 (just a year ago).  </li></ul></li>\n  </ul>\n</blockquote>\n\n<p>In this competition, the principal, \"Trust Your CV\", was also true, as the Public LB was no use. So, the validation strategy was so critical that I want to know how other people took the approach. (However the luck might be most important. Is it true or not? )</p>\n\n<p>Thanks for reading and welcome any comments!</p>",
  "messages": [
    {
      "id": "476303",
      "postDate": "02/21/2019 23:59:21",
      "content": "<p>I briefly write down my solution, which got medal-border score. </p>\n\n<p>So, this is not a complex approach.</p>\n\n<blockquote>\n  <ul>\n  <li><strong>Feature Engineering</strong>: based on <a href=\"https://www.kaggle.com/yoshoku/gacrp-v2-starter-kit\">this kernel</a></li>\n  <li><strong>Modeling...Classification =&gt; Regression</strong>: As discussed <a href=\"https://www.kaggle.com/c/ga-customer-revenue-prediction/discussion/71427\">here</a>, after classifying whether the re-visited purchase will occur or not, I made a regression model which predicts how much the amount of the purchase is.</li>\n  <li><strong>Task Timeframe</strong>: In this task, we have to predict the amount of the purchase from 2018/12/01 to 2019/1/31 per user who has visited between 2018/5/1 and 2018/10/15 with the data whose timeframe is from 2016/4/1 to 2018/4/30 and from 2018/5/1 to 2018/10/15.</li>\n  <li><strong>Validation Strategy</strong>: To consider the problem simply, aiming to ignore the seasonality, I decided to take the following validation strategy: <br>\n  <ul><li>The task is to predict whether re-visited purchase will occur or not between 2018/12/1 and 2019/1/31 by using data whose timeframe is from 2018/5/1 to 2018/10/15.  </li>\n  <li>For validation, I predicted the probability of the occurrance of re-visit between 2017/12/1 and 2018/1/31 by using data whose timeframe is from 2017/5/1 to 2017/10/15 (just a year ago).  </li></ul></li>\n  </ul>\n</blockquote>\n\n<p>In this competition, the principal, \"Trust Your CV\", was also true, as the Public LB was no use. So, the validation strategy was so critical that I want to know how other people took the approach. (However the luck might be most important. Is it true or not? )</p>\n\n<p>Thanks for reading and welcome any comments!</p>",
      "rawMarkdown": "I briefly write down my solution, which got medal-border score. \n\nSo, this is not a complex approach.\n\n&gt;- **Feature Engineering**: based on [this kernel](https://www.kaggle.com/yoshoku/gacrp-v2-starter-kit)\n- **Modeling...Classification =&gt; Regression**: As discussed [here](https://www.kaggle.com/c/ga-customer-revenue-prediction/discussion/71427), after classifying whether the re-visited purchase will occur or not, I made a regression model which predicts how much the amount of the purchase is.\n- **Task Timeframe**: In this task, we have to predict the amount of the purchase from 2018/12/01 to 2019/1/31 per user who has visited between 2018/5/1 and 2018/10/15 with the data whose timeframe is from 2016/4/1 to 2018/4/30 and from 2018/5/1 to 2018/10/15.\n- **Validation Strategy**: To consider the problem simply, aiming to ignore the seasonality, I decided to take the following validation strategy:  \n  - The task is to predict whether re-visited purchase will occur or not between 2018/12/1 and 2019/1/31 by using data whose timeframe is from 2018/5/1 to 2018/10/15.  \n  - For validation, I predicted the probability of the occurrance of re-visit between 2017/12/1 and 2018/1/31 by using data whose timeframe is from 2017/5/1 to 2017/10/15 (just a year ago).  \n    \nIn this competition, the principal, \"Trust Your CV\", was also true, as the Public LB was no use. So, the validation strategy was so critical that I want to know how other people took the approach. (However the luck might be most important. Is it true or not? )\n\nThanks for reading and welcome any comments!",
      "votes": null
    },
    {
      "id": "476309",
      "postDate": "02/22/2019 00:24:38",
      "content": "<p>Congaratularions on medal, nice to see some ml based strategy worked here</p>",
      "rawMarkdown": "Congaratularions on medal, nice to see some ml based strategy worked here",
      "votes": null
    },
    {
      "id": "476322",
      "postDate": "02/22/2019 01:01:52",
      "content": "<p>Thanks, Oleg! There must be other better ml based strategies, which are expected to be shared.</p>",
      "rawMarkdown": "Thanks, Oleg! There must be other better ml based strategies, which are expected to be shared.",
      "votes": null
    }
  ],
  "comments": [
    {
      "id": 476309,
      "author_name": "yaroshevskiy",
      "author_url": "",
      "post_date": "02/22/2019 00:24:38",
      "content": "<p>Congaratularions on medal, nice to see some ml based strategy worked here</p>",
      "votes": null,
      "replies": [
        {
          "id": 476322,
          "author_name": "takaishikawa",
          "author_url": "",
          "post_date": "02/22/2019 01:01:52",
          "content": "<p>Thanks, Oleg! There must be other better ml based strategies, which are expected to be shared.</p>",
          "votes": null,
          "replies": []
        }
      ]
    }
  ],
  "raw_markdown_by_id": {
    "476303": "I briefly write down my solution, which got medal-border score. \n\nSo, this is not a complex approach.\n\n&gt;- **Feature Engineering**: based on [this kernel](https://www.kaggle.com/yoshoku/gacrp-v2-starter-kit)\n- **Modeling...Classification =&gt; Regression**: As discussed [here](https://www.kaggle.com/c/ga-customer-revenue-prediction/discussion/71427), after classifying whether the re-visited purchase will occur or not, I made a regression model which predicts how much the amount of the purchase is.\n- **Task Timeframe**: In this task, we have to predict the amount of the purchase from 2018/12/01 to 2019/1/31 per user who has visited between 2018/5/1 and 2018/10/15 with the data whose timeframe is from 2016/4/1 to 2018/4/30 and from 2018/5/1 to 2018/10/15.\n- **Validation Strategy**: To consider the problem simply, aiming to ignore the seasonality, I decided to take the following validation strategy:  \n  - The task is to predict whether re-visited purchase will occur or not between 2018/12/1 and 2019/1/31 by using data whose timeframe is from 2018/5/1 to 2018/10/15.  \n  - For validation, I predicted the probability of the occurrance of re-visit between 2017/12/1 and 2018/1/31 by using data whose timeframe is from 2017/5/1 to 2017/10/15 (just a year ago).  \n    \nIn this competition, the principal, \"Trust Your CV\", was also true, as the Public LB was no use. So, the validation strategy was so critical that I want to know how other people took the approach. (However the luck might be most important. Is it true or not? )\n\nThanks for reading and welcome any comments!",
    "476309": "Congaratularions on medal, nice to see some ml based strategy worked here",
    "476322": "Thanks, Oleg! There must be other better ml based strategies, which are expected to be shared."
  },
  "source": "meta"
}