student_poisson_mixture

The code is released for the paper Understanding Student Procrastination vis Mixture Models, Educational Data Mining 2018.

Jihyun Park ([email protected])
July 2018

Required Packages

Written in Python2.7.
Python packages numpy, scipy, random, and matplotlib are needed to run the code.

Data

• test_data.csv: Sample data (simulated data) to fit the Poisson mixture model. Each row in the file is considered as a daily activity count vector for a student. 400 rows exist in this sample data.

Demo iPython Notebook

• demo.ipynb: A quick tutorial of using the code.

Code

• pmm.py: Code for fitting Poisson mixture model given a count matrix. The file has two classes--
  PoissonMixture for the model and PoisMixResult for storing and plotting the result.
• utils.py: Has helper functions for calculating log probabilities.