1
$\begingroup$

Lets look for example, at the forecast the sales of a retail outlet.

If I understood the concept correctly, than a lagged feature would be the sales of a previous month t−1.

Would it make sense/is it common practice to create a lagged feature of a feature? For example number of customers and number of customers of a previous month t-1.

I'd be worried by doing so, that I'd give too much weight to unimportant features.

$\endgroup$

1 Answer 1

1
$\begingroup$

Lagged values of features make sense with time series data, this is usually fundamental in time series analysis (because of autocorrelation). Now, whether you should include a lag or not of a feature is a different question, one that is very much data and model dependent, so we cannot answer this definitively.

One thing you might check is the aforementioned autocorrelation, if a feature has autocorrelation then maybe you should include lags, if however there is no autocorrelation then lags are probably useless.

$\endgroup$
2
  • $\begingroup$ Alright, so I'd have to look at the partial auto correlation plots for each feature as well? $\endgroup$ Mar 17 at 13:37
  • $\begingroup$ @RomeroAzzalini It would help. $\endgroup$ Mar 18 at 7:03

Your Answer

By clicking “Post Your Answer”, you agree to our terms of service, privacy policy and cookie policy

Not the answer you're looking for? Browse other questions tagged or ask your own question.