Python Machine Learning

CHAPTER 10Unsupervised Learning—Clustering Using K‐Means

What Is Unsupervised Learning?

So far, all of the machine learning algorithms that you have seen are supervised learning. That is, the datasets have all been labeled, classified, or categorized. Datasets that have been labeled are known as labeled data, while datasets that have not been labeled are known as unlabeled data. Figure 10.1 shows an example of labeled data.

Figure 10.1: Labeled data

Based on the size of the house and the year in which it was built, you have the price at which the house was sold. The selling price of the house is the label, and your machine learning model can be trained to give the estimated worth of the house based on its size and the year in which it was built.

Unlabeled data, on the other hand, is data without label(s). For example, Figure 10.2 shows a dataset containing a group of people's waist circumference and corresponding leg length. Given this set of data, you can try to cluster them into groups based on the waist circumference and leg length, and from there you can figure out the average dimension in each group. This would be useful for clothing manufacturers to tailor different sizes of clothing to fit its customers.

Figure 10.2: Unlabeled data

Unsupervised Learning Using K‐Means ...