Unsupervised learning differs from supervised learning in one crucial way, supervised learning learns from our labelled data whereas unsupervised learning does not have any labelled data and its job is to find something interesting about the data or to find a pattern in the data.
Clustering is one form of unsupervised learning. For example, giving an unsupervised learning model data from a restaurant. The unsupervised learning model can find patterns in orders such as common pairings of foot items or common pairings of drinks with food items.
Clustering could also identify types of customers, whether they be one-time customers, returning customers or infrequent customers.
The data for unsupervised learning will only come with inputs or x but will not have any output or y labels. The model will aim to find structure in the data.
Besides clustering, unsupervised learning can be used for anomaly detection and dimensionality reduction which is a much bigger topic I will cover at a later stage.
