Clustering metrics

Measuring the performance of a clustering algorithm is a little trickier than classification or regression, because clustering is unsupervised machine learning. Thankfully, scikit-learn comes equipped to help us with this as well in a very straightforward manner.

Getting ready

To measure clustering performance, start by loading the iris dataset. We will relabel the iris flowers as two types: type 0 is whenever the target is 0 and type 1 is when the target is 1 or 2:

from sklearn.datasets import load_iris
import numpy as np

iris = load_iris()
X = iris.data
y = np.where(iris.target == 0,0,1)

How to do it...

1. Instantiate a k-means algorithm and train it. Since the algorithm is a clustering one, do not use the target in the training:

from sklearn.cluster import KMeans

kmeans = KMeans(n_clusters=2,random_state=0)
kmeans.fit(X)

2. Now import everything necessary to score k-means through cross-validation. We will use the adjusted_rand_score clustering performance metric:

from sklearn.metrics...