3
$\begingroup$

I read few articles on random forest and its implementation in various fields. But I hardly found any literature on its implementation on forecasting univariate time series.

  • Can it be used for forecasting univariate time series?
  • If yes, how?
  • If no, why?
$\endgroup$
  • 1
    $\begingroup$ It really depends on the nature of your time series, and what your 'modelling', i.e. what you feed to the random forest. If you think finance, I would daresay no. These are at a first and good approximation random walks... If the time series describe a very deterministic phenomenon, it could be. $\endgroup$ – mic Apr 25 '16 at 12:27
  • $\begingroup$ @mic: Actually I was searching for literature related to Finance. Thanks for your response. Cleared one of my confusions. But can it be used if I am trying to classify the trading signals based on derivatives of stock price data i.e 3-days MA, RSI etc? $\endgroup$ – Abhay Bhadani Apr 25 '16 at 17:21
  • $\begingroup$ It would be nice if you could comment on use of bagging for univariate time series forecasting in context of Finance. $\endgroup$ – Abhay Bhadani Apr 25 '16 at 17:31