This repo should be considered as a simple illustration of a training session on a toy dataset to visualize what would be an optimal methodology to :
- Quick transformation of numerical features
- Select an algorithm by cross-validating the test
- Reduce de dimensionnality of a dataset
- Hypertuning a Linear Discrimant Analysis
This work can be looked at a good way to improve ML practices, nothing more.