{
  "id": 27574,
  "title": "evaluation function code?",
  "url": "/competitions/dstl-satellite-imagery-feature-detection/discussion/27574",
  "author_name": "",
  "post_date": "2017-01-11T04:30:25.557Z",
  "votes": 2,
  "comment_count": 3,
  "views": 344,
  "content": "<p>Is there a way to get access to the code that performs evaluation at Kaggle?</p>\n\n<p>I would love to estimate prediction on the hold out set in terms of polygons vs polygons in the same way as it is evaluated at the LB</p>",
  "messages": [
    {
      "id": "155418",
      "postDate": "01/11/2017 04:30:25",
      "content": "<p>Is there a way to get access to the code that performs evaluation at Kaggle?</p>\n\n<p>I would love to estimate prediction on the hold out set in terms of polygons vs polygons in the same way as it is evaluated at the LB</p>",
      "rawMarkdown": "Is there a way to get access to the code that performs evaluation at Kaggle?\r\n\r\nI would love to estimate prediction on the hold out set in terms of polygons vs polygons in the same way as it is evaluated at the LB",
      "votes": null
    },
    {
      "id": "155597",
      "postDate": "01/12/2017 00:19:23",
      "content": "<p>If you are working in R, the <code>rgeos</code> package has the the <code>gArea</code>, <code>gIntersection</code>,  and <code>gUnion</code> functions. Those work on <code>sp::SpatialPolygons*</code>. Then each term in the evaluation metric is just <code>gArea(gIntersection(true, pred)) / gArea(gUnion(true, pred))</code></p>",
      "rawMarkdown": "If you are working in R, the `rgeos` package has the the `gArea`, `gIntersection`,  and `gUnion` functions. Those work on `sp::SpatialPolygons*`. Then each term in the evaluation metric is just `gArea(gIntersection(true, pred)) / gArea(gUnion(true, pred))`",
      "votes": null
    },
    {
      "id": "155709",
      "postDate": "01/12/2017 15:35:23",
      "content": "<p>The production code (in C#) can be found <a href=\"https://gist.github.com/wendykan/2fcbbf95945faa0f2c89895694069010\">here</a> </p>",
      "rawMarkdown": "The production code (in C#) can be found [here][1] \n\n  [1]: https://gist.github.com/wendykan/2fcbbf95945faa0f2c89895694069010",
      "votes": null
    },
    {
      "id": "155774",
      "postDate": "01/12/2017 19:44:37",
      "content": "<p>[quote=Wendy Kan;155709]</p>\n\n<p>The production code (in C#) can be found <a href=\"https://gist.github.com/wendykan/2fcbbf95945faa0f2c89895694069010\">here</a> </p>\n\n<p>[/quote]</p>\n\n<p>Thank you. Are there any <strong>C# / python</strong> experts who can translate the code to python?</p>",
      "rawMarkdown": "[quote=Wendy Kan;155709]\r\n\r\nThe production code (in C#) can be found [here][1] \r\n\r\n  [1]: https://gist.github.com/wendykan/2fcbbf95945faa0f2c89895694069010\r\n\r\n[/quote]\r\n\r\nThank you. Are there any **C# / python** experts who can translate the code to python?",
      "votes": null
    }
  ],
  "comments": [
    {
      "id": 155597,
      "author_name": "davidthaler",
      "author_url": "",
      "post_date": "01/12/2017 00:19:23",
      "content": "<p>If you are working in R, the <code>rgeos</code> package has the the <code>gArea</code>, <code>gIntersection</code>,  and <code>gUnion</code> functions. Those work on <code>sp::SpatialPolygons*</code>. Then each term in the evaluation metric is just <code>gArea(gIntersection(true, pred)) / gArea(gUnion(true, pred))</code></p>",
      "votes": null,
      "replies": []
    },
    {
      "id": 155709,
      "author_name": "wendykan",
      "author_url": "",
      "post_date": "01/12/2017 15:35:23",
      "content": "<p>The production code (in C#) can be found <a href=\"https://gist.github.com/wendykan/2fcbbf95945faa0f2c89895694069010\">here</a> </p>",
      "votes": null,
      "replies": [
        {
          "id": 155774,
          "author_name": "iglovikov",
          "author_url": "",
          "post_date": "01/12/2017 19:44:37",
          "content": "<p>[quote=Wendy Kan;155709]</p>\n\n<p>The production code (in C#) can be found <a href=\"https://gist.github.com/wendykan/2fcbbf95945faa0f2c89895694069010\">here</a> </p>\n\n<p>[/quote]</p>\n\n<p>Thank you. Are there any <strong>C# / python</strong> experts who can translate the code to python?</p>",
          "votes": null,
          "replies": []
        }
      ]
    }
  ],
  "raw_markdown_by_id": {
    "155418": "Is there a way to get access to the code that performs evaluation at Kaggle?\r\n\r\nI would love to estimate prediction on the hold out set in terms of polygons vs polygons in the same way as it is evaluated at the LB",
    "155597": "If you are working in R, the `rgeos` package has the the `gArea`, `gIntersection`,  and `gUnion` functions. Those work on `sp::SpatialPolygons*`. Then each term in the evaluation metric is just `gArea(gIntersection(true, pred)) / gArea(gUnion(true, pred))`",
    "155709": "The production code (in C#) can be found [here][1] \n\n  [1]: https://gist.github.com/wendykan/2fcbbf95945faa0f2c89895694069010",
    "155774": "[quote=Wendy Kan;155709]\r\n\r\nThe production code (in C#) can be found [here][1] \r\n\r\n  [1]: https://gist.github.com/wendykan/2fcbbf95945faa0f2c89895694069010\r\n\r\n[/quote]\r\n\r\nThank you. Are there any **C# / python** experts who can translate the code to python?"
  },
  "source": "meta"
}