{
  "id": 13238,
  "title": "Vowpal Wabbit raw predictions (-r option) to submission file",
  "url": "/competitions/malware-classification/discussion/13238",
  "author_name": "",
  "post_date": "2015-04-05T01:14:25.400Z",
  "votes": null,
  "comment_count": 3,
  "views": 1273,
  "content": "<p>Hi,</p>\n<p>Can you please take a look and suggest what I'm doing wrong?</p>\n<p>I've tried to use Vowpal Wabbit like the following:</p>\n<p>1) Validation:<br>&gt;&gt; vw train -f model --oaa 9 -c --passes 100 -b 24 -l 0.7 --loss_function logistic<br>...<br>passes used = 12<br>...<br>average loss = 0.010129 h</p>\n<p>2) Train:<br>&gt;&gt; vw train -f model --oaa 9 -c --passes 12 -b 24 -l 0.7 --loss_function logistic --holdout_off <br>...</p>\n<p>3) Predict:<br>&gt;&gt; vw test -t -i model -p vw.pred -r vw.rawp</p>\n<p>4) Submission:<br>Convert the vw.rawp file to kaggle submission&nbsp;file by softmax on the raw predictions.</p>\n<p>But I'm getting very bad LB score (~0.08) compared to the validation score.</p>\n<p>Can you please point me to what I'm doing wrong?</p>\n<p>Thanks,<br>C</p>",
  "messages": [
    {
      "id": "69683",
      "postDate": "04/05/2015 01:14:25",
      "content": "<p>Hi,</p>\n<p>Can you please take a look and suggest what I'm doing wrong?</p>\n<p>I've tried to use Vowpal Wabbit like the following:</p>\n<p>1) Validation:<br>&gt;&gt; vw train -f model --oaa 9 -c --passes 100 -b 24 -l 0.7 --loss_function logistic<br>...<br>passes used = 12<br>...<br>average loss = 0.010129 h</p>\n<p>2) Train:<br>&gt;&gt; vw train -f model --oaa 9 -c --passes 12 -b 24 -l 0.7 --loss_function logistic --holdout_off <br>...</p>\n<p>3) Predict:<br>&gt;&gt; vw test -t -i model -p vw.pred -r vw.rawp</p>\n<p>4) Submission:<br>Convert the vw.rawp file to kaggle submission&nbsp;file by softmax on the raw predictions.</p>\n<p>But I'm getting very bad LB score (~0.08) compared to the validation score.</p>\n<p>Can you please point me to what I'm doing wrong?</p>\n<p>Thanks,<br>C</p>",
      "rawMarkdown": "",
      "votes": null
    },
    {
      "id": "69820",
      "postDate": "04/06/2015 19:30:23",
      "content": "<p>Did you shuffled the trainset?</p>",
      "rawMarkdown": "",
      "votes": null
    },
    {
      "id": "69853",
      "postDate": "04/07/2015 01:54:34",
      "content": "<p>No.</p>\n<p>Thanks!</p>\n<p>I'll try that.</p>",
      "rawMarkdown": "",
      "votes": null
    },
    {
      "id": "69859",
      "postDate": "04/07/2015 06:24:16",
      "content": "<p>It looks like that even I didn't shuffled the trainset, it&nbsp;wasn't ordered in any particular order so I've got similar result again.&nbsp;</p>",
      "rawMarkdown": "",
      "votes": null
    }
  ],
  "comments": [
    {
      "id": 69820,
      "author_name": "titericz",
      "author_url": "",
      "post_date": "04/06/2015 19:30:23",
      "content": "<p>Did you shuffled the trainset?</p>",
      "votes": null,
      "replies": []
    },
    {
      "id": 69853,
      "author_name": "clustifier",
      "author_url": "",
      "post_date": "04/07/2015 01:54:34",
      "content": "<p>No.</p>\n<p>Thanks!</p>\n<p>I'll try that.</p>",
      "votes": null,
      "replies": []
    },
    {
      "id": 69859,
      "author_name": "clustifier",
      "author_url": "",
      "post_date": "04/07/2015 06:24:16",
      "content": "<p>It looks like that even I didn't shuffled the trainset, it&nbsp;wasn't ordered in any particular order so I've got similar result again.&nbsp;</p>",
      "votes": null,
      "replies": []
    }
  ],
  "raw_markdown_by_id": {
    "69683": "",
    "69820": "",
    "69853": "",
    "69859": ""
  },
  "source": "meta"
}