Is there an algorithm or modelling technique you find particularly useful? What is it, and why?

  • 16 May 2022
  • 1 reply

Userlevel 3
Badge +1
  • Active Contributor
  • 10 replies

It’s always interesting and insightful to learn how other people like think about solving data science problems, and a big part of this is the tools we like to use.

This thread is for a discussion of useful models, algorithms and techniques -- whatever they may be, and whatever they may be for.

1 reply

Userlevel 3
Badge +1

I can start things off with a super basic but highly underrated (or taken for granted) technique: the log transform. 

From a modelling perspective, taking the log of the independent variable(s) can (sometimes) make a model much better behaved in a variety of different ways. The top answer to this question on cross validated does an excellent job of explaining how.

Log transforms are especially worth considering for multiplicatively scaled signals (such as  highly seasonal timeseries), ratios, and probabilities.