So far, every model has learned from examples with known answers. Unsupervised learning works without labels: it looks for structure in the data on its own, such as natural groups or the most important directions of variation.
You'll start with clustering, using k-means to split data into a chosen number of groups and hierarchical clustering to build a tree of nested groups. Then you'll learn principal component analysis (PCA), which compresses many features into a few while keeping as much information as possible.
Watch the videos in order, then cluster and visualize a dataset of your own, and try PCA to reduce it to two dimensions. When you're done, continue on to Tuning, Interpretation & Pitfalls.
How k-means groups data into k clusters, and how an elbow plot helps you choose a value for k.
How hierarchical clustering builds a tree of nested groups and how to read it as a heatmap and dendrogram.
The main ideas behind PCA: finding the directions that capture the most variation in your data.
How PCA works in detail, from centering the data to finding principal components and choosing how many to keep.