Description
See [this document](../Practicum.md) for general information about the practicums.
Learning objectives:
– Implementing K-Means and Bisecting K-Means clustering algorithms
– Implementing Hierarchical Agglomerative Clustering using different cluster proximities
– Visualizing clusters (scatterplots and dendrograms)
Task 1. Implementing K-Means clustering
– A set of 2D data points are given (generated artificially)
– Select the centroids initially randomly from the data points
– Repeat until the cluster assignments change for less than 1% of the data points
– Visualize the cluster assignments and centroids after each iteration
Task 2. Implementing Bisecting K-Means clustering
– Solve the previous task using the bisecting variant of K-Means
– Measure the quality of the resulting clustering in terms of Sum of Squared Error (SSE)
* How does it compare to the SSE using regular K-Means?
* How does it compare to the SSE of the ‘true’ clustering?
Task 3. Implementing Hierarchical Agglomerative Clustering
– Cluster the “Italian cities” dataset (from the lecture) using Hierarchical Agglomerative Clustering
– Implement the Single link (MIN), Complete link (MAX), and Group average methods for comparing cluster proximities
– Bonus: visualize the different clusterings using dendrograms
