demo_iris: Fits Gaussian Mixture model and computes the KSD value for...
In MinHyung-Kang/KSD: Goodness-of-Fit Tests using Kernelized Stein Discrepancy

We fit a Gaussian Mixture Model for a given dataset (Fisher's Iris), and we compute the KSD P-value on the hold-out test dataset. User may tune the parameters and observe the change in results. Reports average of p-values obtained during each k-fold. It also plots the contour for each k-fold iteration if only 2 dimensions of data are used. If a vector is specified for nClust, the code tries each element as the number of clusters and reports the optimal parameter by choosing one with highest p-value.

1	demo_iris(cols = c(1, 2), nClust = 3, kfold = 5)

`cols`	: Columns of iris data set to use. If 2 dimensions, plots the contour for each k-fold.
`nClust`	: Number of clusters want to estimate with If vector, use each element as number of clusters and reports the optimal number.
`kfold`	: Number of k to use for k-fold