Good point with a little nit. Usually in data science, you "mine" for features or combination of features. That is you collect as much as possible, then actual

Good point with a little nit. Usually in data science, you "mine" for features or combination of features. That is you collect as much as possible, then actual find the critical features e.g. to construct a random forest/decision tree/ ... Here they are going as you point out the other way. And the more I study U.S. history (and politics) the more racist I find it.