Implemented another layer of detail in my food tracking/analyzing project. Specifically, I went from a single sentence in the specifications document that said "Based on the accumulated data, rank the various eating places from least likely to cause bad reactions to most likely to cause problems" to a completed table with counts obtained from sample meals in the document and a worked-out Bayes Theorem equation that gets as specific as:

P(no reaction | ate at the Castle in the Air) = p(...| ...)p(...|...)...etc.
p(...|...) = such-and-such-sum / such-and-such count
p(...|...) = ...
and so on.

Next up: plug in #s, work through 4 more sample analysis questions, and then implement the whole lot in software.