Section 5 – Clustering and Dimensionality Reduction with Unsupervised Learning
The last two chapters of this book examines unsupervised learning models. These are models where there is no target to predict. Even without a target there are many insights that can be gleaned from our data. Dimension reduction with principal component analysis (PCA) allows us to capture the variance of our features with fewer components than the original number of features.
The components created with PCA can be used for visualizations, or to identify processes that are important but cannot really be captured well by each feature. PCA can also be used when we need to reduce the feature space in a supervised learning model. We will demonstrate how to create and evaluate a PCA in the next chapter.
Clustering helps us group instances by those which have more in common with each other than with those in any other group. This often reveals relationships that are not otherwise obvious. We look at...
