Closed ghost closed 9 months ago
Thanks! The iris notebook is meant to showcase the model on simple 2D data. This lets us visualize the resulting model decisions (see Visualizing the Decision Boundary
in the notebook). Therefore, we will keep the 2D projection on purpose by throwing away the other features that result in a more accurate model.
Hi, Thanks for the library. I made a small change in the iris classification demo, using the full dataset (i.e. num_features=4) and then scaling the dataset using sklearn.preprocessing import StandardScaler, the classifier now achieves 100% training accuracy and 98% test accuracy!
Here's the code: