Where to Add Apply Forecast Operator in Random Forest Process

SkyTraderSkyTrader Member Posts: 88 Contributor II
edited August 2020 in Help
Hi there, I set up a Random Forest (RF) for a Time Series prediction and wondered how to add an Apply Forecast operator to get future predictions for my process? Pls see image.



Do I add it to the top row between the RF and Apply Model operators?

Cheers for any help, it'll be interesting to see how this RF compares with the ARIMA model in terms of accuracy of prediction with the real values.


Answers

  • mschmitzmschmitz Administrator, Moderator, Employee, RapidMiner Certified Analyst, RapidMiner Certified Expert, University Professor Posts: 3,233 RM Data Scientist
    Hi,
    Apply forecast is used for specific time series forecasting algorithms (Arima, Holt-Winters).
    Apply model is used for general ML models, which can also be used for forecasting.

    Best,
    Martin
    - Head of Data Science Services at RapidMiner -
    Dortmund, Germany
  • SkyTraderSkyTrader Member Posts: 88 Contributor II
    edited August 2020
    Hi Martin, @mschmitz
    Thanks, I already have 2 x Apply Model as can maybe be seen in the screenshot (if its large enough to see?) and I'm still not sure what to do to get it to predict future dates because the settings for Apply Model all refer to legacy controls that aren't needed!? Please see images -- one blown up larger.

    Will there be a graph for the predictions and actual close in the results and statistics like my ARIMA set up? ...Which I can't see yet with my Random Forests process because the forecast / Apply Model isn't set up correctly?

    Cheers,




  • SkyTraderSkyTrader Member Posts: 88 Contributor II
    Hi Scott @sgenzer, Wondering if you could figure out how to get a future prediction using the Apply Model operator? 
    Thanks! 
  • SkyTraderSkyTrader Member Posts: 88 Contributor II
    edited September 2020
    Cheers Martin, @mschmitz,

    That worked, however my Performance (2) now gives a Relative Error of 0.77% which sounds too good to be true because it seems like a result that would be associated with training and not testing? 

    I have this Random Forest Process (pls see image below) but I cannot get it to predict beyond the last date of my Excel file (11th Sept 2020) using Apply Model, is it correctly connected to the other operators, (it gives no warnings)?







    Many thanks,
Sign In or Register to comment.