Inspired by numerous examples given in scikit-learn documentation and the helper module provided in a fantastic MOOC, the implemented module enables the visualisation of predicted class probabilities for a data set with more than two classes. For instance, the figures below illustrate a synthesised data set containing samples with four classes; the predicted probability contours are obtained with sklearn.neighbors.KNeighborsClassifier and sklearn.ensemble.RandomForestClassifier trained on the two numerical features. In addition, replot method can be passed to ipywidgets.interact function in a Jupyter IDLE, allowing to adjust the model's hyperparameters and immediately observe the changes in contour.
- Edit documentation and type hinting,
- Allow for a contour plot customisation,
- Implement a GUI with streamlit.