{
  "id": 157238,
  "title": "Adversarial Validation 2 - Metadata and Image Size",
  "url": "/competitions/siim-isic-melanoma-classification/discussion/157238",
  "author_name": "Bojan Tunguz",
  "post_date": "2020-06-09T21:19:55.973000",
  "votes": 4,
  "comment_count": 2,
  "views": 0,
  "content": "<p>Turns out that you can get an AUC of 0.7 with just the metadata and image size data. So be careful when using those in your modeling.</p>\n\n<p><a href=\"https://www.kaggle.com/tunguz/adversarial-melanoma-2/\">https://www.kaggle.com/tunguz/adversarial-melanoma-2/</a></p>",
  "messages": [
    {
      "id": 879945,
      "postDate": "2020-06-09T21:19:55.973Z",
      "content": "<p>Turns out that you can get an AUC of 0.7 with just the metadata and image size data. So be careful when using those in your modeling.</p>\n\n<p><a href=\"https://www.kaggle.com/tunguz/adversarial-melanoma-2/\">https://www.kaggle.com/tunguz/adversarial-melanoma-2/</a></p>",
      "rawMarkdown": "Turns out that you can get an AUC of 0.7 with just the metadata and image size data. So be careful when using those in your modeling.\n\nhttps://www.kaggle.com/tunguz/adversarial-melanoma-2/",
      "votes": 4
    },
    {
      "id": 879960,
      "postDate": "2020-06-09T21:36:24Z",
      "content": "<p>I think there is a public which got around .69 with just meta data.  I got .785 with some additional features. You can check it <a href=\"https://www.kaggle.com/awsaf49/xgboost-tabular-data-ml-cv-86-lb-79\">here</a></p>",
      "rawMarkdown": "I think there is a public which got around .69 with just meta data.  I got .785 with some additional features. You can check it [here](https://www.kaggle.com/awsaf49/xgboost-tabular-data-ml-cv-86-lb-79)",
      "replies": [
        {
          "id": 879962,
          "postDate": "2020-06-09T21:37:50.457Z",
          "rawMarkdown": "",
          "isDeleted": true
        }
      ]
    }
  ],
  "comments": [
    {
      "id": 879960,
      "author_name": "Awsaf",
      "author_url": "",
      "post_date": "2020-06-09T21:36:24",
      "content": "<p>I think there is a public which got around .69 with just meta data.  I got .785 with some additional features. You can check it <a href=\"https://www.kaggle.com/awsaf49/xgboost-tabular-data-ml-cv-86-lb-79\">here</a></p>",
      "votes": 0,
      "replies": [
        {
          "id": 879962,
          "author_name": "",
          "author_url": "",
          "post_date": "2020-06-09T21:37:50.457000",
          "content": "",
          "votes": 0,
          "replies": []
        }
      ]
    }
  ],
  "raw_markdown_by_id": {
    "879945": "Turns out that you can get an AUC of 0.7 with just the metadata and image size data. So be careful when using those in your modeling.\n\nhttps://www.kaggle.com/tunguz/adversarial-melanoma-2/",
    "879960": "I think there is a public which got around .69 with just meta data.  I got .785 with some additional features. You can check it [here](https://www.kaggle.com/awsaf49/xgboost-tabular-data-ml-cv-86-lb-79)"
  }
}