Forecasting Next Month data

dgodin3dgodin3 Member Posts: 1 Newbie
edited December 2018 in Help
Hello, I am new to RapidMiner. I have a date set that has monthly values for the last 10 years. The dataset ends in October 2018. How can you predict the value for November 2018? For example,
9/1/2018 =30
10/1/2018 =88
11/1/2018 =??
Tagged:

Answers

  • lionelderkrikorlionelderkrikor Moderator, RapidMiner Certified Analyst, Member Posts: 952   Unicorn
    Hi @dgodin3,

    You have a time-series module in RapidMiner with operators performing predictions.
    For example, you can take a look at the tutorial of the Holt-Winters operator and 
    adapt it to your data.

    Regards,

    Lionel

    NB : It seems to me that you will need of RapidMiner 9.1 (beta) to use the Holt-Winters operator.
    BalazsBaranysgenzerrfuentealba
  • hughesfleming68hughesfleming68 Member Posts: 263   Unicorn
    edited December 2018
    You can also do this with any of the operators that support regression. This includes neural nets, SVM, GLM and Gradient Boosted Trees. The best tool for the job is data dependent. If your data has trend and seasonal components then the new ARIMA operators can be useful. Understand the windowing operator and do the time series tutorials. Those will cover the basics.I would start with the linear predictors and work from there as they are easier to understand.
    BalazsBaranysgenzerrfuentealbaPradeep_K
  • Telcontar120Telcontar120 Moderator, RapidMiner Certified Analyst, RapidMiner Certified Expert, Member Posts: 1,312   Unicorn
    I would probably start with the ARIMA model operator from the new time series extension and the Apply Forecast operator as well, and see what that looks like.  Then you could consider either Holt-Winters or more complex algorithms as well.
    Brian T.
    Lindon Ventures 
    Data Science Consulting from Certified RapidMiner Experts
    sgenzerrfuentealba
  • PapadPapad Computer Science Student Member Posts: 68  Guru
    Hi @dgodin3 ,
    seems like linear regression, deep learning would be a good approach.
    Although ARIMA model is very good about time series predictions so you can take a look at first how forecast validation operator works and after use inside it the arima model operator.
    Hope it helps.
Sign In or Register to comment.