Independent variable classification based on Kano model

Hi everyone,

In a project, there have 1 dependent variable (Y), 20-50 independent variables (Xs), respectively. It is believed these independent variables have linear or no-linear relationships with Y. These conditions are pretty much similiar the idea in Kano model. Now the target is to find a right regression model or other method to sort these varaibles out.

More details: basically, these independent variables have the following approximate relationship with Y, we believe!

1. attractive independent variables: Non-linear, they have significant effect on y when its value over some threshold value, like the function of y=exp(x);

2. performance independent variables: linear, like the function of y=a+bx

3. Must-be variable: non-linear, they have significatin effect on y when its value under some threshold value, like the function of y=ln(x).

Can someone have some ideas about some models and programs classify these independent variables? Thanks.

bond115