Bayesian Gaussian Mixture Model Clustering of the Iris dataset
The Iris dataset contains 3 groups of flowers, with 4 features desribing each flower.
first let's look at the data (4d data so 1d is removed using PCA):
My model, reduces the dimension 4d -> 1d and performs bayesian gmm on the 1d data.
This gives as a 95% accuracy clustering function.
by giving a gaussian over the weights, centered at the PCA 1st component, the model simulates the data nicely, as shown in the folowing p-valuse diagram:
