Split Dataset Features

[ArtNd]

As for the exercise 2 of Lab 1, we need to compute impurity for each candidate split value we want to test.
We then want to attribute each value $\mathcal{X_i}$ to the left or right group.

Then, using a greedy algorithm we want to evaluate all splits and keep the one with the lowest Gini index.

• Split the dataset depending on the candidate value
• Try each candidate value with the splitted dataset and Gini index