Fuzzy-clustering-with-EM(Expectation Maximization)-algorithm

This is the programming assignment of Data Mining course(course code:CSIT5210/MSBD5002) in CSE, HKUST 2017 Fall.

Prerequisites

• Python3+
• Numpy

How to run ?

python3 fcem.py

Problem description

Based on the clickstream event frequency pattern in Q2Q3_input.csv for a given lecture video, apply EM algorithm to cluster the students into two classes with the following initial settings:

• Initial centers: c1 =(1,1,1,1,1,1), c2 = (0,0,0,0,0,0)
• Cluster features: frequency patterns for 6 given clickstream events: load_video,pause_video,play_video,seek_video, speed_change_video and stop_video, you can find them in Q2Q3_input.csv. You need to:
  • (a). Report the updated centers and SSE for the first two iterations.
  • (b). Report the overall iteration step when your algorithm terminates
  • (c). Report the final converged centers for each cluster.

Please use the terminate condition below:

• The sum of L1-distance for each pair of old-new center is smaller than 0.001, or
• The iteration step is greater than the maxinum iteration step 50.