Demo Details¶

The repository contains demo python scripts named demo/demo_1.py and demo/demo_2.py that demonstrate two different uses of the software package.

Demo 1¶

The demo shows an implementation of EM algorithm-based cluster parameter and order estimation for Gaussian mixture model followed by unsupervised classification of datapoints from different clusters.

Steps

First, generate 500 observations from a Gaussian mixture model with 3 clusters.
Then implement “estimate_gm_params” function on the data for the estimation.
Then use “split_classes” followed by “compute_class_likelihood” functions to classify observations from different clusters.

Results

Demo 2¶

The demo shows an implementation of the EM algorithm to estimate the orders and parameters of 2 different Gaussian Mixture models and perform binary maximum likelihood classification.

Steps

First, generate data from 2 Gaussian mixture model each with 3 clusters. The generated data includes training dataset from both mixtures and a combined testing dataset.
Then implement “estimate_gm_params” function on both the training datasets for the estimation.
Finally, use “compute_class_likelihood” function to get the likelihood value to classify testing dataset.

Results