Monday, April 23, 2018

Is bad MLR any better than bad cubic equations?

10 comments:

  1. You've excluded the answer I would have gone with ("depends on your data").

    ReplyDelete
  2. Only one in five knows what mlr is! I am surprised!

    ReplyDelete
  3. Machine Learning in R? Medical Loss Ratio? Major League Rugby?

    (I did eventually figure out you meant multiple linear regression)

    ReplyDelete
  4. Ah. Too many possible acronyms

    ReplyDelete
  5. I suppose, from a theortical standpoint, to do time series analysis?

    Both are crap. I know that. I feel like multiple linear is less crap, but I haven't yet proven that to myself.

    ReplyDelete
  6. MLR will definitely help you convince yourself that you’re seeing real trends, rather than just overfitting your data.

    ReplyDelete
  7. Tony Lower-Basch .... That is just overfitting by another name, right?

    ReplyDelete
  8. William Nichols: Yeah, but a tool that lets you fool yourself about that is ... better? Probably? Maybe?

    ReplyDelete