1. This repo used LDA as a way to discover the most important features https://github.com/ilamaheshwari/LinearDiscriminantAnalysys/blob/master/lda.py 2. QDA and RBF Kernel SVM can be more accurate, could those be included in the notebook?