"Split validation and linear regression"
Alex_Kraiser
MemberPosts:1
Contributor I
hi! tnx for the great software.
i'm trying to learn a multilinear model after feature subset selection using a CFS filter. To this purpose i choosed the operators split validation, optimize selection [trainning], linear regression [training], apply model [testing], performance(regression) [testing], performance(CFS) [optimize selection - evaluation].
Everything seems fine but i noticed that the output of the mod in the split validation (it's equal to output of the linear regression) is different from the output and input of the mod in the operator apply model. without connecting the mod (split validation) to the output then the linear regression and the apply model outputs are the same (equal to the apply model of the previous case).
Could someone please explain this? what is the effect of the output mod in split validation?
tnx in advance, probably i'm missing something obvious (i'm a newbie of data mining and rapidminer).
i'm trying to learn a multilinear model after feature subset selection using a CFS filter. To this purpose i choosed the operators split validation, optimize selection [trainning], linear regression [training], apply model [testing], performance(regression) [testing], performance(CFS) [optimize selection - evaluation].
Everything seems fine but i noticed that the output of the mod in the split validation (it's equal to output of the linear regression) is different from the output and input of the mod in the operator apply model. without connecting the mod (split validation) to the output then the linear regression and the apply model outputs are the same (equal to the apply model of the previous case).
Could someone please explain this? what is the effect of the output mod in split validation?
tnx in advance, probably i'm missing something obvious (i'm a newbie of data mining and rapidminer).
Tagged:
0
Answers
it would really help us, if you could a process that illustrates this behavior
Greetings,
Sebastian