{
  "id": 338620,
  "title": "train labels",
  "url": "/competitions/amex-default-prediction/discussion/338620",
  "author_name": "",
  "post_date": "2022-07-21T07:10:56.260109200Z",
  "votes": 2,
  "comment_count": 2,
  "views": 0,
  "content": "<p>Hi,</p>\n<p>to my surprise I found 118,828 times a 1 in train_labels.csv.<br>\nWith the total amount of customers in the training dataset being 458,913 this mean that 25% of AMEX customers default.<br>\nI expected the percentage to be much lower. What am I missing?</p>\n<p>thanks<br>\nNorbert</p>",
  "messages": [
    {
      "id": "1864596",
      "postDate": "07/21/2022 07:10:56",
      "content": "<p>Hi,</p>\n<p>to my surprise I found 118,828 times a 1 in train_labels.csv.<br>\nWith the total amount of customers in the training dataset being 458,913 this mean that 25% of AMEX customers default.<br>\nI expected the percentage to be much lower. What am I missing?</p>\n<p>thanks<br>\nNorbert</p>",
      "rawMarkdown": "Hi,\n\nto my surprise I found 118,828 times a 1 in train_labels.csv.\nWith the total amount of customers in the training dataset being 458,913 this mean that 25% of AMEX customers default.\nI expected the percentage to be much lower. What am I missing?\n\nthanks\nNorbert",
      "votes": null
    },
    {
      "id": "1864599",
      "postDate": "07/21/2022 07:15:28",
      "content": "<p>Only approximately 1 in 61 customers default. The negative group (non-defaults) was subsampled ~20x or else we would have had a 60:1 imbalance in targets. It is all explained in the <a href=\"https://www.kaggle.com/competitions/amex-default-prediction/overview\" target=\"_blank\"><strong>Overview</strong></a> section.</p>",
      "rawMarkdown": "Only approximately 1 in 61 customers default. The negative group (non-defaults) was subsampled ~20x or else we would have had a 60:1 imbalance in targets. It is all explained in the [**Overview**](https://www.kaggle.com/competitions/amex-default-prediction/overview) section.",
      "votes": null
    },
    {
      "id": "1865953",
      "postDate": "07/22/2022 07:22:02",
      "content": "<p>thanks. Now I understand what the weight of 20 to adjust for downsampling means</p>",
      "rawMarkdown": "thanks. Now I understand what the weight of 20 to adjust for downsampling means",
      "votes": null
    }
  ],
  "comments": [
    {
      "id": 1864599,
      "author_name": "tilii7",
      "author_url": "",
      "post_date": "07/21/2022 07:15:28",
      "content": "<p>Only approximately 1 in 61 customers default. The negative group (non-defaults) was subsampled ~20x or else we would have had a 60:1 imbalance in targets. It is all explained in the <a href=\"https://www.kaggle.com/competitions/amex-default-prediction/overview\" target=\"_blank\"><strong>Overview</strong></a> section.</p>",
      "votes": null,
      "replies": []
    },
    {
      "id": 1865953,
      "author_name": "nhenseler",
      "author_url": "",
      "post_date": "07/22/2022 07:22:02",
      "content": "<p>thanks. Now I understand what the weight of 20 to adjust for downsampling means</p>",
      "votes": null,
      "replies": []
    }
  ],
  "raw_markdown_by_id": {
    "1864596": "Hi,\n\nto my surprise I found 118,828 times a 1 in train_labels.csv.\nWith the total amount of customers in the training dataset being 458,913 this mean that 25% of AMEX customers default.\nI expected the percentage to be much lower. What am I missing?\n\nthanks\nNorbert",
    "1864599": "Only approximately 1 in 61 customers default. The negative group (non-defaults) was subsampled ~20x or else we would have had a 60:1 imbalance in targets. It is all explained in the [**Overview**](https://www.kaggle.com/competitions/amex-default-prediction/overview) section.",
    "1865953": "thanks. Now I understand what the weight of 20 to adjust for downsampling means"
  },
  "source": "meta"
}