DataGenerator Criteria for target function

Answers

• Options

  awchisholm RapidMiner Certified Expert, Member Posts: 458 Unicorn

  April 2011

  Hello,
  
  I had the same question and decided to look at the code to get the answer. To save you some time, you could look here
  
  http://rapidminernotes.blogspot.com/2009/11/examplesetgenerator-parameters.html
  
  
  regards
  
  Andrew

  0
• Options

  gustavo_baldani Member Posts: 3 Contributor I

  April 2011

  Andrew,
  
  thks for replying.
  
  I have seen that post, and it was very usefull. But...  Help me a little....
  
  simple polynomial classification: label = "positive" if att1^4 > 100 otherwise "negative"
  
  2 points:
  
  - Its always that condition? I couldn´t find a relatsionship between my values and that binomial label (In my run).
  - Where can i see that criteria "Here is a list (taken from the source code) " ? In wich part have you seen that?
  
  Sorry for this basic question.
  
  I have been working in BI the last 6 years, bue i´m new in DM and in RapidMiner.
  
  I really like it.
  
  Thks again.

  0
• Options

  awchisholm RapidMiner Certified Expert, Member Posts: 458 Unicorn

  April 2011

  Hello
  
  Royal wedding here - not invited but mildly drunk anyway
  
  I found this file

  C:\Program Files\Rapid-I\RapidMiner5\src\com\rapidminer\operator\generator\SimplePolynomialClassificationFunction.java

  and this code

  
  public class SimplePolynomialClassificationFunction extends ClassificationFunction {
  
  	public double calculate(double[] att) throws FunctionException {
  		if (att.length < 1)
  			throw new FunctionException("Simple polynomial classification function", "needs at least one attribute!");
  		if ((att[0] * att[0] * att[0] * att[0]) > 100)
  			return getLabel().getMapping().mapString("positive");
  		else
  			return getLabel().getMapping().mapString("negative");
  	}
  }
  

  cheers!
  
  Andrew

  0
• Options

  gustavo_baldani Member Posts: 3 Contributor I

  May 2011

  Now yes!!
  
  Thks a lot Andrew!
  
  Best regards.

  0