I am conducting a utility billing analysis with a large number of cases. Each case will have its own regression equation for pre/post weather normalized results. Energy consumption is the dependent variable with weather factors being the basic dependent variables. Homes respond to weather in different ways. Weather normalization is done with heating degree days. Using a fixed reference temperature to determine degree days is one approach, but a better one is to let the reference temperature vary for each model (some homes start heating at 60F, others at 65 or other temps).
In Excel, this can be handled fairly easily using the solver function. Basically, we want to allow for HDD and CDD to float, with solver selecting values (within a reasonable range) that maximize the R-squared. SPSS is far superior to Excel at handling and manipulating large and complex datasets. Is there an equivalent solver function or related technique that may be available in SPSS? We have experienced SPSS users here at work, but none that have tackled this specific issue. The general technique is the Princeton Scorekeeping Method (PRISM) - I know it can programmed in STATA, which is what makes me think SPSS should have this ability, too.
Any pointers are appreciated!