topic Re: Define axis intercept for regression in Discussions

Define axis intercept for regression

LawOfWolf466 — Wed, 15 Nov 2023 16:55:15 GMT

Hello,

Is it possible to specify the intercept for a linear regression under Fit Model?

This means that JMP does not freely select the value for "a" in the linear function y= a+b*x1+c*x2+.... but sets it to 1, for example.

Thanks.

Re: Define axis intercept for regression

Jordan_Hiller — Wed, 15 Nov 2023 19:04:37 GMT

Here's a thought: Rearrange the regression equation.

y-a = b*x1+c*x2+...

This suggests that you can create a new formula data column that subtracts off your desired intercept a from the response variable y. Then run fit model on the new column using the "No Intercept" option (effectively setting the intercept to zero). You can re-add the desired intercept to the resulting prediction expression.

Re: Define axis intercept for regression

Mark_Bailey — Wed, 15 Nov 2023 19:39:56 GMT

You can also specify such a model using the Nonlinear platform, but it is not as convenient as using Fit Least Squares with @Jordan_Hiller's trick.