{
  "fetched_at": "2026-10-19T18:52:52Z",
  "source_url": "https://www.kaggle.com/competitions/aicore-classification/data",
  "pages": [
    {
      "name": "data-description",
      "content": "# Data\n\n__Dataset is secret, you may try to find it's origins and improve your score, but this might be really hard!__\nWe have provided `3` datasets saved as `np.array`s (see [`np.load`](https://numpy.org/doc/stable/reference/generated/numpy.load.html) in order to read in the data):\n\n- `X_train.npy` - training features, shape: `(10000, 100)`\n- `y_train.npy` - training labels, shape: `(10000,)`\n- `X_test.npy` - training features, shape: `(40000, 100)`\n\nYou have to submit prediction on `X_test` as described in __Overview__.\n",
      "isPublished": false,
      "hasIsPublished": false,
      "postTitle": "",
      "hasPostTitle": false,
      "mimeType": "",
      "hasMimeType": false
    },
    {
      "name": "Description",
      "content": "# AiCore classification\n\nWelcome to first AiCore Kaggle InClass competition where you can test your machine learning knowledge!\n\n## General\n\n- __Use any algorithm you want (including, but not limited to ML & DL)__\n- __All we care about is your submission score (but people with best results will be asked for tips and what helped/didn't help__\n\n## Task\n\nYour task is a simple __multiclass classification__ (`10` classes) with limited number of samples (see `Data`).\n- Train on `X_train` & `y_train` labels\n- Predict on `X_test`\n\n## Sending submission\n\n__Provided you have `y_test` as `np.array` variable__ you can use following code to create `submission.csv` for upload (it will be created in your current working directory:\n\n```\npd.DataFrame({\"Id\": np.arange(len(y_test)), \"Category\": y_test}).astype(int).to_csv(\n    \"solution.csv\", index=False\n)\n```\n\nIn general, two columns are needed: \"Id\" (zero-numbered) and \"Category\" containing your predictions.\n\n__Single best submission will be taken at the end and scored against private dataset (or you can choose one manually, it might perform better!)__\n\n## Evaluation\n\nYour submission will be evaluated using plain and simple `accuracy` metric.\n\nYou can use provided metrics by `sklearn` and/or `PyTorch-Lightning` to check your model's performance.\n\n__Remember to use validation set (or more sophisticated methods) to assess model's performance__\n\n## Private vs Public leaderboard\n\n- Public leaderboard is scored against `10000` samples from `X_test`. This is the one you see and can evaluate/compare yourself to others.\n- Private leaderboard contains the rest (`30000`). Results from this evaluation will be publicized at the end of competition __and will be your final place in the leaderboard__.\n\n__If you are lower on the public leaderboard, it doesn't mean you can't jump up on private!__\n\n... but public leaderboard is a good indication how you're performing.\n\n## Baselines\n\nTwo baselines are available:\n- Beat `C` baseline to pass for grad C\n- Beat `B` baseline to pass for grad B\n\nHigher grades will be based on your position on the __private leaderboard__.\n\n## Tips\n\n- Please notice there is __more testing samples than training samples__ (watch out for overfitting!)\n- You have `20` submissions each day, a lot of space for experimentation!\n- You can team in groups of `3` or less, __it won't affect the final grade__",
      "isPublished": false,
      "hasIsPublished": false,
      "postTitle": "",
      "hasPostTitle": false,
      "mimeType": "",
      "hasMimeType": false
    },
    {
      "name": "rules",
      "content": " - Don't cheat (or be smart when doing it)!\n - Be friendly and helpful!\n - Have fun!",
      "isPublished": false,
      "hasIsPublished": false,
      "postTitle": "",
      "hasPostTitle": false,
      "mimeType": "",
      "hasMimeType": false
    }
  ]
}