Predicting how many exceptions of what kind will arise in a given period, so investigation capacity can be staffed rather than reacted to.
Time-series forecasting on aggregated daily counts, with calendar effects such as month-end and holidays as known regressors
pandas rolling averages by day-of-week; statsmodels ARIMA or Holt-Winters; Prophet for calendar effects