A tool for learning mixture models from multimodal continuous data with non-Gaussian modes.
|3 Nov 2011||Ashutosh Tewari||
The project creates a gmcdistribution (Gaussian Mixture Copula distribution) class similar to the gmdistribution (Gaussian Mixture distribution) class of the Matlab's stats toolbox. This class is intended for the characterization of multimodal datasets with non-Gaussian modes. Currently it implements the following methods: 1. fit (for fitting the distribution) 2. cluster (for clustering the datapoints given the learnt model). 3. random (for randomly sampling from the model) 4. pdf 5. cdf