{
  "id": 4655,
  "title": "levels(labels)[rf$test$predicted]",
  "url": "/competitions/digit-recognizer/discussion/4655",
  "author_name": "ecorocks",
  "post_date": "2013-05-21T22:34:17.373000",
  "votes": 0,
  "comment_count": 1,
  "views": null,
  "content": "<p>I found this bit of code in the randomForest benchmark, and was wondering if anyone could explain it? I understand what it does, but why?</p>\r\n<p>For context:</p>\r\n<p><span>labels</span> &lt;- as.factor(<span>train</span>[,<span>1</span>])</p>\r\n<p>train<span> &lt;- </span>train<span>[,</span><span>-1</span><span>]</span></p>\r\n<p></p>\r\n<p>rf<span> &lt;- randomForest(</span>train<span>, </span>labels<span>, </span>xtest<span>=</span>test<span>,\r\n</span>ntree<span>=</span><span>1000</span><span>)</span></p>\r\n<p>predictions<span> &lt;-<strong> levels(</strong></span><strong>labels)[rf$test$predicted]</strong></p>",
  "messages": [
    {
      "id": 25446,
      "postDate": "2013-06-12T14:29:10.133Z",
      "content": "<p>It's basically because the prediction results in labels in the range 1..10 but the output wants the corresponding figures 0..9. So it's basically a typecasting thing. In the kNN benchmark something similar is performed in a different way: predictions&lt;-(0:9)[predictionVector]</p>"
    },
    {
      "id": 24636,
      "postDate": "2013-05-21T22:34:17.373Z",
      "content": "<p>I found this bit of code in the randomForest benchmark, and was wondering if anyone could explain it? I understand what it does, but why?</p>\r\n<p>For context:</p>\r\n<p><span>labels</span> &lt;- as.factor(<span>train</span>[,<span>1</span>])</p>\r\n<p>train<span> &lt;- </span>train<span>[,</span><span>-1</span><span>]</span></p>\r\n<p></p>\r\n<p>rf<span> &lt;- randomForest(</span>train<span>, </span>labels<span>, </span>xtest<span>=</span>test<span>,\r\n</span>ntree<span>=</span><span>1000</span><span>)</span></p>\r\n<p>predictions<span> &lt;-<strong> levels(</strong></span><strong>labels)[rf$test$predicted]</strong></p>"
    }
  ],
  "comments": [
    {
      "id": 25446,
      "author_name": "Frank Inklaar",
      "author_url": "",
      "post_date": "2013-06-12T14:29:10.133000",
      "content": "<p>It's basically because the prediction results in labels in the range 1..10 but the output wants the corresponding figures 0..9. So it's basically a typecasting thing. In the kNN benchmark something similar is performed in a different way: predictions&lt;-(0:9)[predictionVector]</p>",
      "votes": 0,
      "replies": []
    }
  ],
  "raw_markdown_by_id": {
    "25446": "",
    "24636": ""
  }
}