The code in
generative_model.r contains a few primary functions that are very useful for general purposes:
generative_model - This function runs a generative model on a data set, using a combination of numeric and/or categorical variables, with unsupervised, semi-supervised, and supervised modes, depending on the parameters provided and the data set supplied. Its history can be recorded, which allows you to visualize each iteration of its learning.
generative.deviance - This function evaluates the cohesiveness of the model in terms of deviance. Out-of-cluster deviance needs to be implemented to get a good idea of goodness-of-fit for the data.
generative.classify - This takes a model generated by
generative_model and a data set, and then performs classifies the clusters of the data set based on the model's parameters.
base.animated.multiplot - This function creates an animated plot of a model's clustering progress over time. If you are on Windows, you may need to install
animation from my forked repo if your ImageMagick location is in a directory containing a space.
melt.history - A function that expands the clustering/classification history of all the points from a model.
Here is a blog post using this code (and made by the .Rhtml file): http://maxcandocia.com/article/2018/Jan/09/generative-clustering/