Linear combination of attributes

Legacy User

Hello,

I'm using RapidMiner for just a few days and have (propably) a simple question for you: in my data-set I would like to check possible linear combinations of (2 or more) attributes which gives me then the best prediction of the label. In my simple test-case (see below) the factors for att_1 and att_2 are 100 and -100: "label = att_1*100 - att_2*100".

att_1 att_2 label
0.142 0.127 1.500
0.118 0.105 1.300
0.147 0.132 1.500
0.164 0.147 1.500
0.164 0.163 0.000
0.114 0.109 0.500
0.195 0.180 1.500
0.195 0.180 1.500
0.128 0.104 2.500
0.188 0.172 1.500

How can RapidMiner find these factors for me?

Thanks for your help.

Best Regards

Volker

TobiasMalbrecht

Hello Volker,

that task can easily be achieved with a linear regression. Simply use the operator [tt]LinearRegression[/tt] in RapidMiner.

Hope that helps,
Tobias