0

I am new In data mining and Weka. I am working on "Ta_Feng" data set and my Intention is to apply Pearosn's Correlation coefficient to calculate correlation between User and Item attribute in my case UserID, and ProductId are the attributes and both of them are numeric where productID is class(Numeric class) but is there any ways to calculate pearson's correlation between numeric attribute and numeric Class?

Note: my purpose is to use Item-Item (Model based) pearson's correlation coefficient.

Thanks in advance

iforcebd
  • 1
  • 4
  • Questions about how to use software are generally off topic here. This seems to be only about how to use Weka. For the statistical question about how to correlate categorical & continuous variables, see: [Correlations with categorical variables](http://stats.stackexchange.com/q/108007/7290). – gung - Reinstate Monica Jan 15 '17 at 22:31

1 Answers1

0

Well I got the solution,some beginner might have encountered same problem. "correlationAtrributeEval" in waka compute pearson's correlation and class attribute should not be continues value even though it is numeric, what does that mean is- in my case I want to compute pearson's correlation coefficient between two numeric attributes"CustomerID" and "ProductID" my class attribute is "ProductID" and it is numeric attribute but not continues how ? ok say we have 3 products 2034, 2036, and 2036 , now it needs to tell weka that its not continuse attribute though it's numeric. So how to tell that is in-" .arff" file header declare all 3 Products id as Nominal as bellow -

@attribute CustomerID       NUMERIC
@attribute ProductID        {2034,2035,2036}
@attribute Quantity         NUMERIC
@attribute Cost             NUMERIC
@attribute sales            NUMERIC
iforcebd
  • 1
  • 4