Added Scaler in the preprocessing. This has given me better visualizations and seems logical, otherwise PCA/ICA/... is used on unscaled data.
Added possibility to train the visualized algorithm on either the reduced data or the original data (which is better).
Added visualization of the target labels.
Added possibility to select features to be viewed and will generate subplots with them.
Added titles and legends for the mentioned subplots. Legend will be a colorbar for non-boolean features.
A minor issue that I couldn't solve is that the selected features become a list of lists instead of a list of strings during the parsing. E.g : [['number_standard_sections', 'is_ep_not_in_code_section']] instead of ['number_standard_sections', 'is_ep_not_in_code_section'].
See line 101 in model.py and line 106 in visualization.py
Added the following elements :
A minor issue that I couldn't solve is that the selected features become a list of lists instead of a list of strings during the parsing. E.g :
[['number_standard_sections', 'is_ep_not_in_code_section']]instead of['number_standard_sections', 'is_ep_not_in_code_section']. See line 101 in model.py and line 106 in visualization.py