Constrained linear model

[#1]

How can I constrain the range of coefficients of variables in a multi-linear model? 

[#2]

It is difficult to answer your question without any information about your problem. Generally, though, the fitting platforms in JMP for linear models do not use constraints other than the shrinkage techniques in Analyze > Fit Model > Generalized Regression such as ridge regression or LASSO..

You can use the Nonlinear platform and define the linear model with indicator functions that switch coefficients off (set to 0) if they exceed the desired range. The expression would be of the form beta0 + (lo < beta1 < hi)*beta1*X but I do not know about convergence in such a model.

This example illustrates the approach using weight versus height in the Big Class data set. Define the model as a column formula in a new data column: I wanted to constrain the slope to 3.5-4.0 in this example.

Select Analyze > Specialized Modeling > Nonlinear and complete the launch dialog like this:

Click OK and make sure that all the criteria are correct before clicking Go.

Here is the equivalent fit using Bivariate.