TensorFlow Machine Learning Cookbook - Second Edition

There's more...

For this recipe, we clustered the iris dataset into three groups using TensorFlow. Then, we calculated the percentage of data points that fell into similar groups (89%) and plotted a graph of the resulting k-means groups. Since k-means as a classification algorithm is locally linear (linear separator up close), it will be hard to learn the naturally non-linear boundary between Iris versicolor and Iris virginica. But, one advantage is that the k-means algorithm did not need labeled data at all to perform.