3.7 Conclusion and future work

The data abstraction, tsibble, for representing temporal data, extends the tidy data principles into the time domain. Tidy data takes shape in the realm of time with the new contextual semantics: index and key. The index variable provides direct support to an exhaustive set of ordered objects. The key, which can consist of single or multiple variables, identifies observational units over time. These semantics further determine unique data entries required for a valid tsibble. It shepherds raw temporal data through the tidying stage of an analysis pipeline to the next exploration stage to fluently gain insights.

The supporting toolkits articulate the temporal data pipeline, with the shared goal of reducing the time between framing of data questions and the code realization. The rapid iteration for broader understanding of the data is achieved through frictionlessly shifting among transformation, visualization, and modeling, using the standardized tsibble data infrastructure.

Future work includes allowing user-defined calendars, so that the tsibble structure respects structural missing observations. For example, a call center may operate only between 9:00 am and 5:00 pm on week days, and stock trading resumes on Monday straight after Friday. No data available outside trading hours would be labeled as structural missingness. Customer calendars can be embedded into the tsibble framework in theory. A few R packages provide functionality to create and manage many specific calendars, such as the bizdays package (Freitas 2018) for business days calendars. However, a generic flexible calendar system is lacking, and requires complex implementation, so this is left for future work.