This is the best-written thing I've seen on this extremely maddening topic. It's as if there's some conspiracy to make the use of deep learning unnecessarily complicated and difficult. I wish someone would write a library that facilitates this moving of data between dataframes and keras...

I do have a question / problem.

Line 37 is commented "# how many predictors/Xs/features we have to predict y"

However, giving it any number other than 1 results in a "tuple index out of range" error. I want to feed it more than one input. So, I'd like to request a follow-up that shows how to do that?

Thanks so much!