C2_W4_Lab_02_Tree_Ensemble. Continuous valued features

Among the 5 categorical features in the heart.csv dataset:

there are 3 continuous value features:

The categorical features have been taken care of with one-hot encoding - pd.get_dummies.
The continuous value features are used as is in the model and during the training.

Shouldn’t the continuous value features be split as mentioned in the lecture Continuous valued features? I guess that, for example, the level of cholesterol can have some impact on the hart disease.

Am I missing something, seeing something wrong?


If I am correct the algorithm takes care of the continuous value features when it chooses the splits during training. When choosing a split it will consider categorical and continuous features as described in lectures. Decision trees work for the both types of data, you encode categorical features because most implementations of the model don`t accept non-numerical data as inputs.

Thanks! I was on the wrong path. :crazy_face: