Good write-up, but IMO these kinds of posts are not useful for most people (despite there being many just like it) - the 30,000 foot view is too much information for a beginner but too basic for someone who is knowledgeable in the field. In my ten years of being fairly immersed in machine learning there was maybe a one month period where an article like this would have been very useful, and that was a couple of years in after I had mastered the basics of the more useful techniques (like generalized linear models and SVMs) and was ready to learn what else was out there. Honestly, learning the high-level overview was not hard at that point.
I think a solid machine learning book coupled with a chart like this (http://scikit-learn.org/stable/tutorial/machine_learning_map...) is the most useful way to approach machine learning.