{
  "id": 251054,
  "title": "How will lag variables be applied on hidden private test set?",
  "url": "/competitions/mlb-player-digital-engagement-forecasting/discussion/251054",
  "author_name": "",
  "post_date": "2021-07-05T17:27:52.419174200Z",
  "votes": null,
  "comment_count": 2,
  "views": 0,
  "content": "<p>Hello All,</p>\n<p>I've seen in a few time series competitions in Kaggle that kernels with lag features sometimes do really well. I was hoping to incorporate them in my submissions as well. I've seen a few notebooks for this competition where participants have added lags in the range of 1-30. This works well now since the public test dataset is for month of May 2021 which immediately follows the training dataset. </p>\n<p>But then how are we supposed to keep lag variables when final test set for evaluation will be be for the month of August 2021? Will we get an update to training dataset at some stage where more training data till the month of July is going to be available?</p>",
  "messages": [
    {
      "id": "1377244",
      "postDate": "07/05/2021 17:27:52",
      "content": "<p>Hello All,</p>\n<p>I've seen in a few time series competitions in Kaggle that kernels with lag features sometimes do really well. I was hoping to incorporate them in my submissions as well. I've seen a few notebooks for this competition where participants have added lags in the range of 1-30. This works well now since the public test dataset is for month of May 2021 which immediately follows the training dataset. </p>\n<p>But then how are we supposed to keep lag variables when final test set for evaluation will be be for the month of August 2021? Will we get an update to training dataset at some stage where more training data till the month of July is going to be available?</p>",
      "rawMarkdown": "Hello All,\n\nI've seen in a few time series competitions in Kaggle that kernels with lag features sometimes do really well. I was hoping to incorporate them in my submissions as well. I've seen a few notebooks for this competition where participants have added lags in the range of 1-30. This works well now since the public test dataset is for month of May 2021 which immediately follows the training dataset. \n\nBut then how are we supposed to keep lag variables when final test set for evaluation will be be for the month of August 2021? Will we get an update to training dataset at some stage where more training data till the month of July is going to be available?",
      "votes": null
    },
    {
      "id": "1377292",
      "postDate": "07/05/2021 18:06:37",
      "content": "<p>we will get an update of training dataset by the 20th of July</p>",
      "rawMarkdown": "we will get an update of training dataset by the 20th of July",
      "votes": null
    },
    {
      "id": "1377574",
      "postDate": "07/06/2021 01:04:06",
      "content": "<p>Okay. Thank you!</p>",
      "rawMarkdown": "Okay. Thank you!",
      "votes": null
    }
  ],
  "comments": [
    {
      "id": 1377292,
      "author_name": "l0glikelihood",
      "author_url": "",
      "post_date": "07/05/2021 18:06:37",
      "content": "<p>we will get an update of training dataset by the 20th of July</p>",
      "votes": null,
      "replies": [
        {
          "id": 1377574,
          "author_name": "arpitsolanki14",
          "author_url": "",
          "post_date": "07/06/2021 01:04:06",
          "content": "<p>Okay. Thank you!</p>",
          "votes": null,
          "replies": []
        }
      ]
    }
  ],
  "raw_markdown_by_id": {
    "1377244": "Hello All,\n\nI've seen in a few time series competitions in Kaggle that kernels with lag features sometimes do really well. I was hoping to incorporate them in my submissions as well. I've seen a few notebooks for this competition where participants have added lags in the range of 1-30. This works well now since the public test dataset is for month of May 2021 which immediately follows the training dataset. \n\nBut then how are we supposed to keep lag variables when final test set for evaluation will be be for the month of August 2021? Will we get an update to training dataset at some stage where more training data till the month of July is going to be available?",
    "1377292": "we will get an update of training dataset by the 20th of July",
    "1377574": "Okay. Thank you!"
  },
  "source": "meta"
}