Open
Description
Most signals will use rates rather than counts, month-over-month changes, and differences in rates between performance and comparators. It might make sense to do that work efficiently on the whole performance dataframe. This needs some thought about where it goes in the overall process of grouping by measure, checking validity, and extending the dataframe with different columns.
Also, things like limiting the number of months to twelve, checking measures against the local master list, etc. Some stuff belongs to individual signals, some stuff is global. We might even calculate basic, "core", signals like current and last month gaps, current and last month streaks and trends. This bears discussion.