{
  "id": 81987,
  "title": "Train users and Test users are different",
  "url": "/competitions/ga-customer-revenue-prediction/discussion/81987",
  "author_name": "yukari17",
  "post_date": "2019-02-26T16:43:36.679000",
  "votes": -15,
  "comment_count": 0,
  "views": 0,
  "content": "<p>My team used the probability that y exceeds 0 as part of the final prediction model.  </p>\n\n<blockquote>\n  <h1>Train(My team used)</h1>\n  \n  <p>N : 358288 <br>\n  Predicted purchasers : 110.589(The correct value is 109)  </p>\n  \n  <h1>Test</h1>\n  \n  <p>N : 296530 <br>\n  Predicted purchasers : 387.947(The correct value is ???)  </p>\n  \n  <p>purchaser is y &gt; 0.  </p>\n</blockquote>\n\n<p>If all 0 prediction is scored 0.32 in Train, and RMSE of Train purchasers and Test purchasers is the same, calculate all 0 prediction score in Test,  </p>\n\n<p>MSE(Only Train purchasers) = (0.32^2) * 358288 / 109 <br>\nAll 0 prediction Score in Test = sqrt( MSE(Only Train purchasers) * 387.947 / 296530 ) = 0.6636.  </p>\n\n<p>Actually, MSE of Train purchasers and Test purchasers is different. Therefore, I think that it is no wonder that all 0 prediction is scored 0.88xx.  </p>\n\n<p>I am sorry if it's wrong.  </p>",
  "messages": [
    {
      "id": 478791,
      "postDate": "2019-02-26T16:43:36.680Z",
      "content": "<p>My team used the probability that y exceeds 0 as part of the final prediction model.  </p>\n\n<blockquote>\n  <h1>Train(My team used)</h1>\n  \n  <p>N : 358288 <br>\n  Predicted purchasers : 110.589(The correct value is 109)  </p>\n  \n  <h1>Test</h1>\n  \n  <p>N : 296530 <br>\n  Predicted purchasers : 387.947(The correct value is ???)  </p>\n  \n  <p>purchaser is y &gt; 0.  </p>\n</blockquote>\n\n<p>If all 0 prediction is scored 0.32 in Train, and RMSE of Train purchasers and Test purchasers is the same, calculate all 0 prediction score in Test,  </p>\n\n<p>MSE(Only Train purchasers) = (0.32^2) * 358288 / 109 <br>\nAll 0 prediction Score in Test = sqrt( MSE(Only Train purchasers) * 387.947 / 296530 ) = 0.6636.  </p>\n\n<p>Actually, MSE of Train purchasers and Test purchasers is different. Therefore, I think that it is no wonder that all 0 prediction is scored 0.88xx.  </p>\n\n<p>I am sorry if it's wrong.  </p>",
      "rawMarkdown": "My team used the probability that y exceeds 0 as part of the final prediction model.  \n  \n&gt;#Train(My team used)  \n&gt;N : 358288  \n&gt;Predicted purchasers : 110.589(The correct value is 109)  \n  \n&gt;#Test  \n&gt;N : 296530  \n&gt;Predicted purchasers : 387.947(The correct value is ???)  \n &gt; \n&gt;purchaser is y &gt; 0.  \n  \nIf all 0 prediction is scored 0.32 in Train, and RMSE of Train purchasers and Test purchasers is the same, calculate all 0 prediction score in Test,  \n  \nMSE(Only Train purchasers) = (0.32^2) * 358288 / 109  \nAll 0 prediction Score in Test = sqrt( MSE(Only Train purchasers) * 387.947 / 296530 ) = 0.6636.  \n  \nActually, MSE of Train purchasers and Test purchasers is different. Therefore, I think that it is no wonder that all 0 prediction is scored 0.88xx.  \n  \nI am sorry if it's wrong.  ",
      "votes": -15
    }
  ],
  "comments": [],
  "raw_markdown_by_id": {
    "478791": "My team used the probability that y exceeds 0 as part of the final prediction model.  \n  \n&gt;#Train(My team used)  \n&gt;N : 358288  \n&gt;Predicted purchasers : 110.589(The correct value is 109)  \n  \n&gt;#Test  \n&gt;N : 296530  \n&gt;Predicted purchasers : 387.947(The correct value is ???)  \n &gt; \n&gt;purchaser is y &gt; 0.  \n  \nIf all 0 prediction is scored 0.32 in Train, and RMSE of Train purchasers and Test purchasers is the same, calculate all 0 prediction score in Test,  \n  \nMSE(Only Train purchasers) = (0.32^2) * 358288 / 109  \nAll 0 prediction Score in Test = sqrt( MSE(Only Train purchasers) * 387.947 / 296530 ) = 0.6636.  \n  \nActually, MSE of Train purchasers and Test purchasers is different. Therefore, I think that it is no wonder that all 0 prediction is scored 0.88xx.  \n  \nI am sorry if it's wrong.  "
  }
}