I work for a museum and I've been tasked with building a method to predict our attendance. I've created different multiple regression models for each unique attendance segment (i.e. tourists vs locals). Each model predicts the the weekly attendance for a given segment.

What I'm trying to figure out is, how do I summarize the model predictions both: a) across models for a given week: i.e. I expect that in week 1 we will have 500 local visitors and 1000 tourists for a total of 1500 visitors with a prediction interval of +- z. b) longitudinally within a model: i.e. I expect that we will have 25,000 local visitors this year with a prediction interval of +- z.

I'm asking this independent of software, however, I used MiniTab to create the regression models.