"Date type for SVM"

msaqibmsaqib Member Posts: 2 Contributor I
edited May 23 in Help
I'm trying to build an SVM/RVM based predictor for energy prices. The dataset contains date values and corresponding prices. Of course, SVM can not work with the 1/4/09 date values (polynomial). I applied the Nominal2Date converter and it changed it to a date string. I was still where I started. Then, I applied Date2Numeric, but that only allowed me to extract one of the parts (year, month or day). Any idea how I should get all three separated out?
Thanks and best regards
Tagged:

Answers

  • landland RapidMiner Certified Analyst, RapidMiner Certified Expert, Member Posts: 2,525   Unicorn
    Hi,
    that's quite easy: Add the operator three times with different parameter settings.
    By the way: If you want to predict future prices, you will need to "window" your series data, otherwise the SVM can only use the pure date information, and you can imagine, that nobody, not even the smartest, could predict the energy price only from knowing thats the 5th februray of 2011.

    Greetings,
      Sebastian

  • msaqibmsaqib Member Posts: 2 Contributor I
    Actually, I did place three of them consecutively. I used a ExampleSetWriter to see what came out, and it turned out that what I got was only the output of the last Date2Numeric. Being a newbie to RapidMiner might be the reason I don't know how to get all of them.
Sign In or Register to comment.