Competitive regression: for each visit, predict the unbiased (“true”) OFF MDS-UPDRS motor score. Minimize RMSE on the unlabeled test set.
- GUIDED.md - walkthrough: join Kaggle and form a team first, then setup, shared Exploratory Data Analysis (EDA), models, and Kaggle upload
- Kaggle competition: ibm-probabl-hackaton (private)
- CONTEXT.md - scientific background, goals, and modelling pitfalls for the Parkinson's dataset eligibility, and what makes a Submission valid
Every Kaggle upload must put a Skore Hub EstimatorReport URL (https://ibm.skore.probabl.ai/…) in the Submission Description. See the competition rules.