model-v0.1.0

Model 0.1.0

Trained 2026-11-02T17:15:20.319542+00:00 with seed 42; classifier logistic_regression.

Test metrics

• accuracy: 0.826
• macro F1: 0.807
• balanced accuracy: 0.844

class	precision	recall	F1	support
glioma	0.907	0.771	0.834	354
meningioma	0.761	0.678	0.717	273
pituitary	0.941	0.936	0.938	357
no_tumor	0.590	0.992	0.740	125

Below the 0.90 macro-F1 target: the team accepted these metrics as a known limitation for the hackathon demo. Many tumors from the dataset's second source (images that are not 512x512) are called no_tumor (docs/ml-engine/decisions.md, D24, D26 and D28).

Trained and scored on public sample data. Not for clinical use.