Toy Datasets Doubt

#1

In the make classification, why do we have specified both the number of classes and the number of clusters per class?
What do we mean when we have more than one clusters for a class?

#2

no_of_clusters per class by default is 2.
Refer Documentation : http://scikit-learn.org/stable/modules/generated/sklearn.datasets.make_classification.html

Suppose you have two types of apples - red apples and green apples they will form different clusters but if the class is APPLE, they will belong to same class. So they have different values of features -color/taste/sweetness etc. and hence will look as different regions in space(more than one cluster) but the class remains the same APPLE.

1 Like