0
$\begingroup$

I am predicting the binary class, i.e. if it's in top10 or not, of a security based upon it's performance using predictors from current time. So it's simply a cross sectional classifier. As of now I've used random forest and neural net for this purpose.

Now I want to extend it so that I can one step ahead prediction of the class. Please suggest some starting point. I understand it might be open ended question. Thanks for reading.

I know how to use time series, but I'm not sure how to go so for a classifier. Also all the predictors are numeric variables, none of them are categorical.

I'm doing all in R, so it would be great if I get related pointers, not a strict constraint though.

Improve this question
$\endgroup$
2
  • $\begingroup$ You want to add lags of explanatory variables to your model? $\endgroup$
    – snoram
    Commented Mar 13, 2015 at 16:15
  • $\begingroup$ If it works, sure why not. Thing is does ARMA work with Classifier? Initially I thought simply shifting the Class one step ahead. Not sure how it would play out. $\endgroup$
    – fras
    Commented Mar 13, 2015 at 20:34

0

Reset to default

Your Answer

By clicking “Post Your Answer”, you agree to our terms of service and acknowledge you have read our privacy policy.

Browse other questions tagged or ask your own question.