{
  "id": 118363,
  "title": "Data Leakage",
  "url": "/competitions/ashrae-energy-prediction/discussion/118363",
  "author_name": "Naruhiko Nakanishi",
  "post_date": "2019-11-21T02:04:40.024000",
  "votes": 2,
  "comment_count": 6,
  "views": 0,
  "content": "<p><em><strong>Data Leakage</strong></em></p>\n\n<p>Data leakage (or leakage) happens when your training data contains information about the target, but similar data will not be available when the model is used for prediction. This leads to high performance on the training set (and possibly even the validation data), but the model will perform poorly in production.</p>\n\n<p>In other words, leakage causes a model to look accurate until you start making decisions with the model, and then the model becomes very inaccurate.</p>\n\n<p>There are two main types of leakage: target leakage and train-test contamination.</p>\n\n<p><a href=\"https://www.kaggle.com/alexisbcook/data-leakage\">https://www.kaggle.com/alexisbcook/data-leakage</a></p>",
  "messages": [
    {
      "id": 678113,
      "postDate": "2019-11-21T02:04:40.023Z",
      "content": "<p><em><strong>Data Leakage</strong></em></p>\n\n<p>Data leakage (or leakage) happens when your training data contains information about the target, but similar data will not be available when the model is used for prediction. This leads to high performance on the training set (and possibly even the validation data), but the model will perform poorly in production.</p>\n\n<p>In other words, leakage causes a model to look accurate until you start making decisions with the model, and then the model becomes very inaccurate.</p>\n\n<p>There are two main types of leakage: target leakage and train-test contamination.</p>\n\n<p><a href=\"https://www.kaggle.com/alexisbcook/data-leakage\">https://www.kaggle.com/alexisbcook/data-leakage</a></p>",
      "rawMarkdown": "***Data Leakage***\n\nData leakage (or leakage) happens when your training data contains information about the target, but similar data will not be available when the model is used for prediction. This leads to high performance on the training set (and possibly even the validation data), but the model will perform poorly in production.\n\nIn other words, leakage causes a model to look accurate until you start making decisions with the model, and then the model becomes very inaccurate.\n\nThere are two main types of leakage: target leakage and train-test contamination.\n\nhttps://www.kaggle.com/alexisbcook/data-leakage",
      "votes": 2
    },
    {
      "id": 685809,
      "postDate": "2019-12-02T10:38:00.963Z",
      "content": "<p><em><strong>Data Leakage Exercise</strong></em></p>\n\n<p>refer to:\n<a href=\"https://www.kaggle.com/grapestone5321/ashrae-sample-submission-and-data-leakage-exercise\">https://www.kaggle.com/grapestone5321/ashrae-sample-submission-and-data-leakage-exercise</a></p>",
      "rawMarkdown": "***Data Leakage Exercise***\n\nrefer to:\nhttps://www.kaggle.com/grapestone5321/ashrae-sample-submission-and-data-leakage-exercise"
    },
    {
      "id": 685808,
      "postDate": "2019-12-02T10:34:16.177Z",
      "content": "<p><em><strong>Data Leakage in Machine Learning</strong></em></p>\n\n<p>refer to:\n<a href=\"https://machinelearningmastery.com/data-leakage-machine-learning/\">https://machinelearningmastery.com/data-leakage-machine-learning/</a></p>",
      "rawMarkdown": "***Data Leakage in Machine Learning***\n\nrefer to:\nhttps://machinelearningmastery.com/data-leakage-machine-learning/\n\n"
    },
    {
      "id": 685740,
      "postDate": "2019-12-02T07:49:31.223Z",
      "content": "<p>Naruhiko ,  Can you please explain it ( in simple sentence) in the context of this competition.</p>",
      "rawMarkdown": "Naruhiko ,  Can you please explain it ( in simple sentence) in the context of this competition.",
      "replies": [
        {
          "id": 685807,
          "postDate": "2019-12-02T10:32:34.450Z",
          "content": "<p>The goal of predictive modeling is to develop a model that makes accurate predictions on new data, unseen during training.</p>\n\n<p>This is a hard problem, because we cannot evaluate the model on something we don’t have.</p>\n\n<p>Therefore, we must estimate the performance of the model on unseen data by training it on only some of the data we have and evaluating it on the rest of the data.</p>",
          "rawMarkdown": "The goal of predictive modeling is to develop a model that makes accurate predictions on new data, unseen during training.\n\nThis is a hard problem, because we cannot evaluate the model on something we don’t have.\n\nTherefore, we must estimate the performance of the model on unseen data by training it on only some of the data we have and evaluating it on the rest of the data."
        }
      ]
    },
    {
      "id": 687665,
      "postDate": "2019-12-04T16:24:55.587Z",
      "rawMarkdown": "",
      "votes": 1,
      "isDeleted": true,
      "replies": [
        {
          "id": 687777,
          "postDate": "2019-12-04T19:54:23.747Z",
          "content": "<p>Oo...\nmost likely the admin though he got hacked given the sudden scraping activity...</p>",
          "rawMarkdown": "Oo...\nmost likely the admin though he got hacked given the sudden scraping activity...",
          "votes": 1
        }
      ]
    }
  ],
  "comments": [
    {
      "id": 685809,
      "author_name": "Naruhiko Nakanishi",
      "author_url": "",
      "post_date": "2019-12-02T10:38:00.963000",
      "content": "<p><em><strong>Data Leakage Exercise</strong></em></p>\n\n<p>refer to:\n<a href=\"https://www.kaggle.com/grapestone5321/ashrae-sample-submission-and-data-leakage-exercise\">https://www.kaggle.com/grapestone5321/ashrae-sample-submission-and-data-leakage-exercise</a></p>",
      "votes": 0,
      "replies": []
    },
    {
      "id": 685808,
      "author_name": "Naruhiko Nakanishi",
      "author_url": "",
      "post_date": "2019-12-02T10:34:16.177000",
      "content": "<p><em><strong>Data Leakage in Machine Learning</strong></em></p>\n\n<p>refer to:\n<a href=\"https://machinelearningmastery.com/data-leakage-machine-learning/\">https://machinelearningmastery.com/data-leakage-machine-learning/</a></p>",
      "votes": 0,
      "replies": []
    },
    {
      "id": 685740,
      "author_name": "yogi",
      "author_url": "",
      "post_date": "2019-12-02T07:49:31.223000",
      "content": "<p>Naruhiko ,  Can you please explain it ( in simple sentence) in the context of this competition.</p>",
      "votes": 0,
      "replies": [
        {
          "id": 685807,
          "author_name": "Naruhiko Nakanishi",
          "author_url": "",
          "post_date": "2019-12-02T10:32:34.450000",
          "content": "<p>The goal of predictive modeling is to develop a model that makes accurate predictions on new data, unseen during training.</p>\n\n<p>This is a hard problem, because we cannot evaluate the model on something we don’t have.</p>\n\n<p>Therefore, we must estimate the performance of the model on unseen data by training it on only some of the data we have and evaluating it on the rest of the data.</p>",
          "votes": 0,
          "replies": []
        }
      ]
    },
    {
      "id": 687665,
      "author_name": "",
      "author_url": "",
      "post_date": "2019-12-04T16:24:55.587000",
      "content": "",
      "votes": 1,
      "replies": [
        {
          "id": 687777,
          "author_name": "eagle4",
          "author_url": "",
          "post_date": "2019-12-04T19:54:23.747000",
          "content": "<p>Oo...\nmost likely the admin though he got hacked given the sudden scraping activity...</p>",
          "votes": 1,
          "replies": []
        }
      ]
    }
  ],
  "raw_markdown_by_id": {
    "678113": "***Data Leakage***\n\nData leakage (or leakage) happens when your training data contains information about the target, but similar data will not be available when the model is used for prediction. This leads to high performance on the training set (and possibly even the validation data), but the model will perform poorly in production.\n\nIn other words, leakage causes a model to look accurate until you start making decisions with the model, and then the model becomes very inaccurate.\n\nThere are two main types of leakage: target leakage and train-test contamination.\n\nhttps://www.kaggle.com/alexisbcook/data-leakage",
    "685809": "***Data Leakage Exercise***\n\nrefer to:\nhttps://www.kaggle.com/grapestone5321/ashrae-sample-submission-and-data-leakage-exercise",
    "685808": "***Data Leakage in Machine Learning***\n\nrefer to:\nhttps://machinelearningmastery.com/data-leakage-machine-learning/\n\n",
    "685740": "Naruhiko ,  Can you please explain it ( in simple sentence) in the context of this competition.",
    "687665": ""
  }
}