In this project, we use differents methods to transform our dataset (usually dimension modification) before making prediction thanks to machine learning and regressions. Those methods are:
- Nothing: No modification
- PCA : We keep as much as variance as possible within two columns
- Regularisation: With a Lasso regretion, we deduce the alpha value
- Variable selection: We keep only columns that contain the more variance
Finally, we compare the results to see which one is truly efficient