Checklist USA-NA-AIO 2026 Round 1 · Task 9
Breast Cancer Classification with k-NN
Train a k-nearest-neighbours pipeline for binary classification of a 30-feature tabular dataset and submit an inference function.
The task
Problem 9 (50 points) provides a training set with 30 numeric features and a binary target (2026_USAAIO_Round1_breast_cancer_train.csv on Hugging Face). A hidden test set is used for grading.
The contestant submits one notebook with preprocessing, model construction, training and a function my_prediction(X_test) that returns a pandas Series of predicted labels; the organisers run the notebook from top to bottom and call the function on the hidden test features. The notebook must end with a text cell summarising the approach, the intuition behind it and alternatives considered.
Abridged by SOTA from the official materials. The official statement has the exact rules, and it wins wherever this summary differs.
At a glance
- You get
- Training CSV with 30 numeric features and a target column.
- You submit
- A Jupyter notebook defining
my_prediction(X_test) -> pandas Series. - Scoring
- Macro-averaged F1 score (f1-macro) on the hidden test set; grading also considers whether the notebook runs end to end and the quality of reasoning.
- Rules
- k-Nearest Neighbours is the only supervised model allowed; any preprocessing or feature engineering is allowed.
- Only scikit-learn, numpy, pandas and matplotlib may be imported; no deep neural networks.
- Format
- 2026 USA-NA-AIO Round 1, 30 January 2026 (date printed on the problem set). Individual, proctored.