ML Unsupervised Learning

For discovering hidden patterns in data. The target variable is not available. Dataset is a collection of unlabelled examples. Usually large, as biggerr samples better reflect the probability distribution of the data.

• Clustering … organize items based on similarity according to a distance metric
• Association … discover co-occurrences between features (if you buy milk, you prob also buy cheese)

Applications:

• Dimensionality Reduction
• Anomaly detection
• Topic Modelling