Time series analysis is quite relevant and important for governmental applications for a number of reasons. First, governments both large and small, are the keepers of some of the most important time series data in the world, including the US jobs report, ocean temperature data (that is, global warming data), and local crime statistics. Second, governments by definition provide some of the most essential services we all rely on, and thus they need to be reasonably adept forecasters of demand if they don’t want to grossly overspend on, or understaff, those services. Thus, all aspects of time series are relevant to government purposes: storage, cleaning, exploration, and forecasting.
As I mentioned back in Chapter 2 when discussing “found” time series, a very high percentage of all government data can look a lot like time series data with some restructuring. Generally, most government data sets are the result of ongoing data collection rather than a single slice of time. However, government data sets can be daunting for a number of reasons:
Inconsistent recordkeeping (due to organizational constraints or political forces changing over time)
Opaque or confusing data practices
Enormous data sets with relatively low information content
Nonetheless, it can be quite interesting to look at government data sets both for intellectual interest and for many practical purposes. In this chapter we explore a governmental data set that consists of all ...