I'm helping my wife try and navigate IBM SPSS and python. She knows SPSS, and I kinda know python -- We might be able to work together. As it stands, I understand that I can call small snippets of python from within an SPSS syntax. While this is useful for looping and conditional branching based on data, it seems a little fuzzy to me. It almost feels like Inversion of Control, but not really.
I was wondering is it possible to have a python script, external to an spss syntax, that can still use the SPSS libraries in any meaningful way, or do I have to keep my scripts confined to the SPSS syntax and runtime?
Yes, you can run Statistics in external mode from a Python or R program. You might have to add the SPSS Python directory to your Python search path, but then just do import spss
and run your Python code. The only thing you can't do is Viewer and user interface stuff, because there is no SPSS UI in that mode. By default, you will get output as text (which you can turn off when you get the hang of things). If you want better quality output, you can use OMS to capture output in a wide variety of formats.
Note that you need a compatible version of Python if you don't use the one installed with SPSS. That would be 2.7 for most Statistics versions. The Python installed with Statistics is not registered, but you can install a standard version from Python.org and just add the SPSS Python directory to the search path.
HTH