CS330 Machine Learning Competition
Predict AAER-related financial reporting from SEC EDGAR fundamentals. Train a classifier on SEC EDGAR fundamentals, upload it with a predict.py script, and get scored automatically on hidden test data.
- Submission deadline
- Dec 11 Dec 11, 2026 11:59 PM PST
- Teams competing
- 0 1 models submitted so far
- Primary metric
- AUPRC AUPRC (average precision) on the hidden public test subset
Current leaderboard
Each team's best public score. Final rankings use a separate private subset.
| Rank | Team | Public score · AUPRC (average precision) |
|---|
No scores yet
The board fills in as soon as the first submission is evaluated.
Updates automatically every 30 seconds
How it works
Four steps from notebook to leaderboard.
- 1
Register
Create your team and save the access token. You'll use it every time you sign in.
- 2
Train
Use
train.csvandvalidation.csvto build a model that predictsAAER_related. - 3
Package
Zip your saved model with a
predict.pythat follows the standard command-line interface. - 4
Submit
Upload the ZIP. It runs in an isolated sandbox and is scored on hidden labels within minutes.
Evaluation rules
- Up to 5 accepted submissions per team in any rolling 24-hour window. Failed evaluations still count; rejected uploads don't.
- One evaluation per team at a time; each run gets 2 CPU cores, 4 GB RAM, 120 seconds and no internet access.
- Scored by AUPRC (average precision). The public board shows your best public score.
- Final ranking: your best public submission is scored on the private subset. Your best private score is never cherry-picked.
Prediction interface
python predict.py \
--input /data/test_features.csv \
--model /submission/model.joblib \
--output /output/predictions.csv
Output: observation_id,predicted_probability, one row per input row, probabilities in [0, 1].