Checklist MAIO 2025 Malaysia AI Olympiad (national competition) · Task 1
Eye for Feature Engineering
Write one engineered feature that lets a fixed logistic-regression classifier separate a peculiar 2D dataset.
The task
A two-feature dataset (feature1, feature2, class) is passed through a scikit-learn LogisticRegression; with the baseline feature the classifier scores 0 precision, recall and F1.
The contestant may only edit the function create_new_feature(X), which adds feature3, and the cell for extra imports; the model and scoring cells in the 'Your submission' section must not be changed, and that section must not depend on the scratchpad code.
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
- A CSV dataset loaded from the organisers' bucket (
maio_2025_eye_for_feature_engineering.csv) and the baseline notebook. - You submit
- The notebook with the edited
create_new_feature() (notebook submission only). - Scoring
- Up to 10 pts = F1 x 10 (training-set F1 of the fixed logistic regression, positive label 1); +3 pts if F1 >= 0.5 without neural networks (learnable weights and biases); +2 pts if F1 >= 0.5 and the %%timeit cell reports <= 10 ms. Partial credit at discretion.
- Rules
- Only
create_new_feature() and the import cell may be edited in the submission section. - Bonus points require no neural networks and a feature runtime of at most 10 ms.
- Only
- Format
- MAIO 2025, the first Malaysia AI Olympiad: online over the weekend of 14-16 March 2025 (48 hours), with objective questions and three handcrafted challenges (maximum total 55 points). This problem was designed to be solved within 2 hours; all LLM assistance, including reasoning models, was allowed. Graded completely by hand (notebook submission); slightly modified for distribution.