In Chapter 3, Supervised Machine Learning Models for Your Data, we focused on supervised machine learning algorithms. This chapter will build on the previous chapters in that we will continue the tour of the machine learning paradigm offered in IBM Cloud. The chapter will cover supervised versus unsupervised as well as semi-supervised learning.

Supervised learning problems are usually categorized into regression and classification problems, and we saw the ways that using IBM Watson Studio and its model builder feature can help solve for those sort of problems.

Unsupervised learning, on the other hand, allows us to approach problems when we might have little or no idea what the results should or would look like. Here, in these types of problems, we can attempt to derive structure from the data itself by clustering (the data) based upon relationships identified among the variables within the data, even if we don't necessarily know the effect of those variables.

This chapter will focus on the concept of unsupervised machine learning and its related topics.

Moreover, this chapter will discuss some common clustering algorithms. And finally, this chapter will conclude by discussing online versus batch learning concepts.

We will divide this chapter into the following areas:

• Unsupervised learning
• Semi-supervised learning
• Anomaly detection
• Online and/or batch learning