"Correlation using group by ."

ratheesan

Hi,
I already posted this in data mining forum.But Mr.Ingo said this is the proper place to post this one ,thats why i am again posting it here.

Hi,
I have three attributes state,height,weight.Here state is string and other two are numeric.My objective is to  calculate correlation between height and weight group by state.ie,in sas\sql its like select  corr(height,weight) group by state.How can I perform this task in RM.

Thanks
Ratheesan.


Hi,

my first idea was of course the aggregation operator but since you are calculating an aggregated value based on two columns for each group-by-value separately, it is probably the best idea to use the operator "Loop Values" for this. This board here is for general discussions about data mining, please post RapidMiner specific questions in the forum board "Data Mining / ETL / BI Processes". There people are more likely to help you with such a concrete problem.

Cheers,
Ingo

But In aggregate operator does not contain correlation.Anybody can suggest me how to apply this in correlation also.

By
Ratheesan

fischer

Hi,

thanks for posting in the right forum. The answer Ingo has given in the other forum, however, is valid also here. Try to use a combination of "Loop Values", "Aggregation", and "Generate Attributes".

Best,
Simon