It's classic Reddit that the first response to this useful post is to assert (a) that the post belongs in another forum, and (b) that 'we know' the stuff presented in the post. I think, though, that this general level guidance is exactly what nose-to-the-algorithm data scientists need to see from time to time. In any case, the post is more generally useful, and offers a good guide to the issues that should be considered when implementing recommendation systems anywhere, including learning analytics. If your project hasn't taken these ten items into consideration, then that's a point of failure.
Today: 5 Total: 108 [Direct link] [Share]
