What to Brew is a crowd sourced database of how well beer styles (Stout, Pale Ale, etc) and additions (chocolate, bacon, cherry) go with each other.
I have a GitHub repo with some sample analysis using Jupyter Notebooks and helpers that should make it easy for you to get started answering some of your own questions. Check it out at wtb-data. If you have some interesting analysis, please contribute!
Feel free to check out the aggregated results by downloading data_export.json. This is a dictionary with additions as keys, and each value is a dictionary with styles as keys, and good/gross as values. It is automatically updated each night, and only includes additions that have had all 96 styles voted on.
Read about some of the ways I've visualized the data here.
You can also explore the data on Datazar.
I'd love to hear how you're using this data- drop me a line at [email protected].