Fall 2010 CSC 466: Knowledge Discovery from Data Alexander Dekhtyar

In this assignment you will use unsupervised learning techniques to discover clusters in a number of simple datasets. You will implement two cluster analysis methods, k-means clustering and agglomerative hierarchical clustering. Assignment Preparation This is a pair programming assignment. Each student teams up with a partner. While this assignment can be easily split into two parts (one per-son does k-means clustering algorithm, the other — hierarchical clustering algorithm), because of timing issues1 and because I want everyone to un-derstand the structure of both algorithms, I strongly recommend that both students take active part in implementing both algorithms. Note: Open-source versions of clustering algorithms are plentiful. It is considered cheating to re-use code for clustering algorithms that was written by someone else. (You may use code from other explicitly acknowledged sources for other purposes in the program, e.g., for parsing input). Data We provide multiple simple datasets for use in this assignment. Each dataset consists of two files: a CSV (comma-separated values) file containing the actual data and a text.txt file containing the header of the dataset, specifying the names of the attributes. 1Hierarchical clustering will be covered later in the course than the k-means clustering algorithm. 1 Most of the datasets were adopted fro