Goodness-of-Fit Tests using Kernelized Stein Discrepancy
Tests 1-dimensional Gaussian Mixture Models.
Tests multidimensional Gaussian Mixture Models.
Fits Gaussian Mixture model and computes the KSD value for the model
Shows KSD p value change with respect variation in noise
Tests 1-dimensional Gamma Distribution with customized parameters
Tests 1-dimensional Gaussian Distribution with customized parameters
Returns a Gaussian Mixture Model
Estimate Kernelized Stein Discrepancy (KSD)
Calculates the likelihood for a given dataset for a GMM
Returns a perturbed model of given GMM
Plots histogram for 1-d GMM given the dataset
Calculates the posterior probability for a given dataset for a GMM
Generates dataset from Gaussian Mixture Model
Score function for given GMM : calculates score function dlogp(x)/dx f...
An adaptation of Kernelized Stein Discrepancy, this package provides a goodness-of-fit test of whether a given i.i.d. sample is drawn from a given distribution. It works for any distribution once its score function (the derivative of log-density) can be provided. This method is based on "A Kernelized Stein Discrepancy for Goodness-of-fit Tests and Model Evaluation" by Liu, Lee, and Jordan, available at <arXiv:1602.03253>.