{
  "id": 347656,
  "title": "Data first, Feature second, Algorithm third",
  "url": "/competitions/amex-default-prediction/discussion/347656",
  "author_name": "",
  "post_date": "2022-08-25T02:10:53.510900300Z",
  "votes": 5,
  "comment_count": 1,
  "views": 0,
  "content": "<p>Congrats to all the winners!</p>\n<p>I think this competition again shows the theorem that data is the most important, then feature, then algorithm. For this competition, the data is sufficient and the feature is already generated and quite quite strong, hence the improving space for algorithm is limited(the result shows that). </p>\n<p>Eager to study from all the guys creativity solutions, NO MATTER how it scored!</p>",
  "messages": [
    {
      "id": "1912852",
      "postDate": "08/25/2022 02:10:53",
      "content": "<p>Congrats to all the winners!</p>\n<p>I think this competition again shows the theorem that data is the most important, then feature, then algorithm. For this competition, the data is sufficient and the feature is already generated and quite quite strong, hence the improving space for algorithm is limited(the result shows that). </p>\n<p>Eager to study from all the guys creativity solutions, NO MATTER how it scored!</p>",
      "rawMarkdown": "Congrats to all the winners!\n\nI think this competition again shows the theorem that data is the most important, then feature, then algorithm. For this competition, the data is sufficient and the feature is already generated and quite quite strong, hence the improving space for algorithm is limited(the result shows that). \n\nEager to study from all the guys creativity solutions, NO MATTER how it scored!",
      "votes": null
    },
    {
      "id": "1912904",
      "postDate": "08/25/2022 03:11:06",
      "content": "<p>I agree with that order.</p>\n<p>I spent most of my time generating features. My different GBT models CV all seemed to be really close, not deserving much of my time.</p>\n<p>I've seen that on public notebooks LGBM was better than XGB but I could never beat my XGB models with LGBM using the feature sets I created… </p>",
      "rawMarkdown": "I agree with that order.\n\nI spent most of my time generating features. My different GBT models CV all seemed to be really close, not deserving much of my time.\n\nI've seen that on public notebooks LGBM was better than XGB but I could never beat my XGB models with LGBM using the feature sets I created...",
      "votes": null
    }
  ],
  "comments": [
    {
      "id": 1912904,
      "author_name": "hinepo",
      "author_url": "",
      "post_date": "08/25/2022 03:11:06",
      "content": "<p>I agree with that order.</p>\n<p>I spent most of my time generating features. My different GBT models CV all seemed to be really close, not deserving much of my time.</p>\n<p>I've seen that on public notebooks LGBM was better than XGB but I could never beat my XGB models with LGBM using the feature sets I created… </p>",
      "votes": null,
      "replies": []
    }
  ],
  "raw_markdown_by_id": {
    "1912852": "Congrats to all the winners!\n\nI think this competition again shows the theorem that data is the most important, then feature, then algorithm. For this competition, the data is sufficient and the feature is already generated and quite quite strong, hence the improving space for algorithm is limited(the result shows that). \n\nEager to study from all the guys creativity solutions, NO MATTER how it scored!",
    "1912904": "I agree with that order.\n\nI spent most of my time generating features. My different GBT models CV all seemed to be really close, not deserving much of my time.\n\nI've seen that on public notebooks LGBM was better than XGB but I could never beat my XGB models with LGBM using the feature sets I created..."
  },
  "source": "meta"
}