Fitted value and confidence interval of part of a regression model

[Samantha St. Amand]

Hello,

I am estimating a non-linear regression of the following form:

nl (y = {alpha} + ({theta} + {beta1}*x1 + {beta2}*x2 + 0.5*x3)*({eta1}*z1 + {eta2}*z2 + {eta3}*z3}))

After estimating the model I would like to generate a fitted value and confidence interval for the first component of the equation, specifically fittedvalue = {theta} + {beta1}*x1 + {beta2}*x2 + 0.5*x3

I am having trouble finding a way to calculate both the fitted value and confidence interval. If there are no canned commands to perform these results, I have no trouble generating the fitted value manually but I do not know how I would generate the confidence interval.

Any help on the matter would be greatly appreciated.

Kindest regards,
Sam

[Jorge Eduardo Perez Perez]

See help predictnl

[Samantha St. Amand]

Hi Jorge,

The problem is that I only want confidence intervals for one part of the model, not the entire predicted model. I'm not sure how I would segregate that portion described above.

Thank you,
Sam

[Samantha St. Amand]

I see now that it is quite simple to specify the model in nlpredict. For anyone looking for a similar solution, the code is:

predictnl fittedvalue = _b[/theta] + _b[/beta1]*x1 + _b[/beta2]*x2, ci(lb ub)