{
  "id": 73323,
  "title": "Which metric is used for evaluation ? Is it rmse or something else?",
  "url": "/competitions/elo-merchant-category-recommendation/discussion/73323",
  "author_name": "",
  "post_date": "2018-12-02T01:37:18.854676900Z",
  "votes": null,
  "comment_count": 4,
  "views": 0,
  "content": "<p>Does anyone have any idea about the metric being used in this competition. </p>\n\n<p>Asking this because the train and validation set rmse is somewhere around 1.6 for my model, but on the leaderboard it's 3.911. </p>\n\n<p>Thanks for the help.</p>\n\n<p>Best,\nHarshit</p>",
  "messages": [
    {
      "id": "431283",
      "postDate": "12/02/2018 01:37:18",
      "content": "<p>Does anyone have any idea about the metric being used in this competition. </p>\n\n<p>Asking this because the train and validation set rmse is somewhere around 1.6 for my model, but on the leaderboard it's 3.911. </p>\n\n<p>Thanks for the help.</p>\n\n<p>Best,\nHarshit</p>",
      "rawMarkdown": "Does anyone have any idea about the metric being used in this competition. \n\nAsking this because the train and validation set rmse is somewhere around 1.6 for my model, but on the leaderboard it's 3.911. \n\nThanks for the help.\n\nBest,\nHarshit",
      "votes": null
    },
    {
      "id": "431315",
      "postDate": "12/02/2018 03:24:14",
      "content": "<p><a href=\"https://www.kaggle.com/c/elo-merchant-category-recommendation#evaluation\">https://www.kaggle.com/c/elo-merchant-category-recommendation#evaluation</a></p>",
      "rawMarkdown": "https://www.kaggle.com/c/elo-merchant-category-recommendation#evaluation",
      "votes": null
    },
    {
      "id": "431338",
      "postDate": "12/02/2018 04:31:59",
      "content": "<p>Thanks !</p>",
      "rawMarkdown": "Thanks !",
      "votes": null
    },
    {
      "id": "431878",
      "postDate": "12/03/2018 03:49:52",
      "content": "<p>Sounds to me like you clipped the outliers of the target or replaced them in some way. RMSE is still correct but there are outliers in the test set so you can't just drop them. </p>",
      "rawMarkdown": "Sounds to me like you clipped the outliers of the target or replaced them in some way. RMSE is still correct but there are outliers in the test set so you can't just drop them.",
      "votes": null
    },
    {
      "id": "431896",
      "postDate": "12/03/2018 04:26:22",
      "content": "<p>Thanks whamp. That's was the case indeed.</p>",
      "rawMarkdown": "Thanks whamp. That's was the case indeed.",
      "votes": null
    }
  ],
  "comments": [
    {
      "id": 431315,
      "author_name": "chambox",
      "author_url": "",
      "post_date": "12/02/2018 03:24:14",
      "content": "<p><a href=\"https://www.kaggle.com/c/elo-merchant-category-recommendation#evaluation\">https://www.kaggle.com/c/elo-merchant-category-recommendation#evaluation</a></p>",
      "votes": null,
      "replies": [
        {
          "id": 431338,
          "author_name": "harshit92",
          "author_url": "",
          "post_date": "12/02/2018 04:31:59",
          "content": "<p>Thanks !</p>",
          "votes": null,
          "replies": []
        }
      ]
    },
    {
      "id": 431878,
      "author_name": "hampsonw",
      "author_url": "",
      "post_date": "12/03/2018 03:49:52",
      "content": "<p>Sounds to me like you clipped the outliers of the target or replaced them in some way. RMSE is still correct but there are outliers in the test set so you can't just drop them. </p>",
      "votes": null,
      "replies": [
        {
          "id": 431896,
          "author_name": "harshit92",
          "author_url": "",
          "post_date": "12/03/2018 04:26:22",
          "content": "<p>Thanks whamp. That's was the case indeed.</p>",
          "votes": null,
          "replies": []
        }
      ]
    }
  ],
  "raw_markdown_by_id": {
    "431283": "Does anyone have any idea about the metric being used in this competition. \n\nAsking this because the train and validation set rmse is somewhere around 1.6 for my model, but on the leaderboard it's 3.911. \n\nThanks for the help.\n\nBest,\nHarshit",
    "431315": "https://www.kaggle.com/c/elo-merchant-category-recommendation#evaluation",
    "431338": "Thanks !",
    "431878": "Sounds to me like you clipped the outliers of the target or replaced them in some way. RMSE is still correct but there are outliers in the test set so you can't just drop them.",
    "431896": "Thanks whamp. That's was the case indeed."
  },
  "source": "meta"
}