{
  "fetched_at": "2026-10-20T02:00:11Z",
  "source_url": "https://www.kaggle.com/competitions/1000-genomes-ancestry/data",
  "pages": [
    {
      "name": "data-description",
      "content": "The majority of the columns in the data represent alleles -- pairs of letters that identify the nucleotides at that position in an individual's genome. The alleles can be any two-letter permutation of letters A,T,C,G. The \"rs*\" identifier in the column name is an identifier that can be used to find more information about the allele variation at that location in the genome. The data from this competition was collected using the [ezancestry](https://github.com/arvkevi/ezancestry) Python library.\n\nThere is also a gender field included in these data.\n\nThere are 55 AISNPs in `kidd_train.csv` and 128 AISNPs in `seldin_train.csv`. Both files have the same individuals.\n\nThe test dataset has missing data for some of the AISNPs.\n\nYou are predicting the superpopulation column.\n\n## Files\n\n*   **kidd_train.csv** - the training set with Kidd AISNPs\n*   **seldin_train.csv** - the training set with Seldin AISNPs\n*   **kidd_test.csv** - the test set with Kidd AISNPs\n*   **seldin_test.csv** - the test set with Seldin AISNPs\n*   **sample_submission.csv** - a sample submission file in the correct format\n\n## Columns\n\n*   `id` - A unique identifier for each individual in the 1000 Genomes Project.\n*   `rs*` - The alleles for each AISNP.\n*   `gender` - The reported gender of the individual in the 1000 Genomes Project.\n*   `superpopulation` - The reported superpopulation of the individual in the 1000 Genomes Project.\n\nsuperpopulation codes:\nAFR: Africa\nEAS: East Asia\nEUR: Europe\nSAS: South Asia\nAMR: Americas",
      "isPublished": false,
      "hasIsPublished": false,
      "postTitle": "",
      "hasPostTitle": false,
      "mimeType": "",
      "hasMimeType": false
    },
    {
      "name": "Evaluation",
      "content": "Submissions are evaluated on [log loss](https://scikit-learn.org/stable/modules/generated/sklearn.metrics.log_loss.html).\n\n## Submission File\nFor each id in the test set, you must predict a probability for each ancestry label. The file should contain a header and have the following format:\n\n|   id |       AFR |      AMR |      EAS |      EUR |       SAS |\n|-----|----------|---------|---------|---------|----------|\n|  689 | 0.177696  | 0.206956 | 0.211084 | 0.247305 | 0.156959  |\n| 1196 | 0.230893  | 0.102298 | 0.24698  | 0.199245 | 0.220584  |\n| 1766 | 0.0837201 | 0.478872 | 0.211402 | 0.156158 | 0.0698465 |\n| 1898 | 0.051935  | 0.261239 | 0.273983 | 0.235532 | 0.177311  |\n| 2071 | 0.0942584 | 0.222179 | 0.136058 | 0.244469 | 0.303035  |\n\nNote that the table is truncated. The submission file should have 1 id column and 5 ancestry columns.",
      "isPublished": false,
      "hasIsPublished": false,
      "postTitle": "",
      "hasPostTitle": false,
      "mimeType": "",
      "hasMimeType": false
    },
    {
      "name": "rules",
      "content": "Don't cheat!",
      "isPublished": false,
      "hasIsPublished": false,
      "postTitle": "",
      "hasPostTitle": false,
      "mimeType": "",
      "hasMimeType": false
    },
    {
      "name": "abstract",
      "content": "Welcome to the 1000 Genomes Project ancestry prediction competition.\n\nGeneticists identified locations in the human genome that can assess an individual's ancestry, called ancestry-informative single nucleotide polymorphisms (AISNPs). These locations, denoted by rsid numbers, display ancestry-specific variation. For example, at location [rs3737576](https://www.ncbi.nlm.nih.gov/snp/rs3737576), 32% of individuals from the 1000 Genomes Project who were assigned American ancestry have a C allele, while the study-wide frequency of the C allele was only 8%.\n\nThere are many locations like this across the human genome. AISNPs from two peer-reviewed publications are available in this competition. [Kidd et al.](https://pubmed.ncbi.nlm.nih.gov/24508742/) identified 55 AISNPs, and [Seldin et al.](https://www.ncbi.nlm.nih.gov/pmc/articles/PMC3073397/) identified 128 AISNPs with varying degrees of discriminatory power. You can learn more about original ancestry classifications from the 1000 Genomes Project [here](https://www.internationalgenome.org/category/population/) and [here](https://www.internationalgenome.org/data-portal/population).\n\nThe dataset in this competition contains alleles from the 1000 Genomes Project for 2,504 individuals across 183 AISNPs.\n\n**Goal:** This competition aims to use AISNPs from the 1000 Genomes Project to predict individuals' ancestry. \nCollaborating with the winner to incorporate the winning model into the [ezancestry](https://github.com/arvkevi/ezancestry) Python library would be fun.\n\nThe term \"ancestry\" can be confusing, especially in genetics. I encourage you to read more about the term [here](https://www.ncbi.nlm.nih.gov/pmc/articles/PMC7082057/).\n\nThis competition is not affiliated with the 1000 Genomes Project.",
      "isPublished": false,
      "hasIsPublished": false,
      "postTitle": "",
      "hasPostTitle": false,
      "mimeType": "",
      "hasMimeType": false
    }
  ]
}