Auto-ML¶
Upload a CSV, pick a target column — Auto-ML trains 5 models and picks the best one. Powered by scikit-learn running in your browser via Pyodide.
Features¶
- Auto-detection — determines classification vs regression from target column
- 5 algorithms compared:
- Classification: Logistic Regression, Random Forest, Gradient Boosting, KNN, SVM
- Regression: Linear Regression, Random Forest, Gradient Boosting, SVR, KNN
- Results table — accuracy, precision, recall, F1 (or R², MAE, MSE)
- Feature importance — bar chart from tree-based models
- Test predictions — enter values, predict with best model
- Export — download best model as pickle
How it works¶
- Drop a CSV file
- Select target column
- Select feature columns (auto-excludes non-numeric)
- Click "Run Auto-ML"
- Wait ~30-120 seconds while 5 models train
- See comparison table with best model highlighted
- Test predictions and export
Technical details¶
- Runs full CPython via Pyodide (WebAssembly)
- scikit-learn, pandas, numpy loaded in-browser
- First load downloads ~15 MB of packages (cached after)
- Handles up to ~100K rows comfortably