RapidMiner

Trouble with Applying a Linear Regression Model with Nominal Attribute

Contributor

Trouble with Applying a Linear Regression Model with Nominal Attribute

Hi there,

I build a linear regression model with nominal attribute converted to numerical attribute (e.g. SMOKE=0, SMOKE=1). I tried both effect encoding and dummy encoding. Then I tried applying this model to a testing dataset in which the value of SMOKE attribute in all observations is 0. Rapidminer says "attributes do not match". How can I solve this?

 

Thanks! 

1 REPLY
Moderator

Re: Trouble with Applying a Linear Regression Model with Nominal Attribute

Hi, you need to apply the preprocessing model you get from your first Nominal to Numerical operator. This will ensure to have the same data structure. Best, Martin
--------------------------------------------------------------------------
Head of Data Science Services at RapidMiner