I enjoyed your post about the LDA (the Dirichlet one; yet to read the Discriminant one). It strikes the right balance between explaining the conceptual aspects of what's going on, and including concrete examples of the data set used and the graphs of the results. Too many articles in this area are either too much into theory crafting with not much practical stuff to prove it, or sometimes go the other way to "here's a code dump, gather whatever you can with it, good luck!".

(One suggestion: sometimes the context of the post isn't actually as obvious as it seems to us. For eg., the post about the Stan Vim plugin makes no mention of what Stan is, and it's a slightly annoying name to DDG properly. A link to the Stan language's site would've saved me - the reader - a minute and some annoyance.)