Tools for scanning for optimal support vector machine classifier parameters
Discussed and demonstrated here.
- scikit-learn
- bokeh
- numpy
- tqdm
Scanning and plotting are accessed through an objected-oriented API, which is demonstrated in demo.ipynb.
- Generalize for other kernel types, and eventually for other kinds of classifiers.
- Add regression analysis to calculate the optimal parameters at each level.
- Automate the iterative "zooming in" process.