5.1 Software development

A particular emphasis of this thesis is on translating research methodologies in the form of open source R packages: sugrrants, tsibble, and mists. Figure 5.1 gives an overview of my Git commits to these repositories, and Figure 5.2 shows the daily downloads of the packages from the RStudio mirror (one of 90 CRAN mirrors) since they were available on CRAN.

Patterns of my package development effort during my PhD years based on Git commits to three repositories, sugrrants, tsibble, and mists. Scatter plots of weekly totals are overlaid with a loess smoother. The sugrrants package was the first project with much initial energy, followed by small constant maintenance. The tsibble package has been a major project with ongoing constant development and bursts of effort in response to users’ feedback. The mists package has been a recent intense project.

Figure 5.1: Patterns of my package development effort during my PhD years based on Git commits to three repositories, sugrrants, tsibble, and mists. Scatter plots of weekly totals are overlaid with a loess smoother. The sugrrants package was the first project with much initial energy, followed by small constant maintenance. The tsibble package has been a major project with ongoing constant development and bursts of effort in response to users’ feedback. The mists package has been a recent intense project.

Impact of these works (sugrrants and tsibble) as measured by daily downloads (on square root scale) from the RStudio mirror since they landed on CRAN. The tsibble package has an increasing trend, suggesting the steady adoption of the new data structure.

Figure 5.2: Impact of these works (sugrrants and tsibble) as measured by daily downloads (on square root scale) from the RStudio mirror since they landed on CRAN. The tsibble package has an increasing trend, suggesting the steady adoption of the new data structure.

5.1.1 sugrrants

The sugrrants package implements the idea of displaying data in the familiar calendar style using frame_calendar() and facet_calendar(). The research article, a shorter version of Chapter 2, has been awarded the best student paper prize from ASA Sections on Statistical Computing and Statistical Graphics and ACEMS Business Analytics in 2018. There has been a grand total of 14,706 downloads from the RStudio mirror dating from 2017-07-28 to 2019-09-20; and it has been starred 48 times on Github so far. The homepage at https://pkg.earo.me/sugrrants contains detailed documentation and a vignette on frame_calendar().

5.1.2 tsibble

The tsibble package provides a data infrastructure and a domain specific language in R for representing and manipulating tidy temporal data. This package provides the fundamental architecture that other temporal tools will be built upon. For example, a new suite of time series analysis packages, titled “tidyverts”, have been developed for the new “tsibble” object. The tsibble package has won the 2019 John Chambers Statistical Software Award from the ASA Sections on Statistical Computing and Statistical Graphics. It has been downloaded 41,058 times from the RStudio mirror since it landed on CRAN; and it has received 241 stars on Github. These metrics are the indicators of my research impact, the recognition by professionals, and the uptakes by users. The website (https://tsibble.tidyverts.org) includes full documentation and three vignettes about the package usage.

5.1.3 mists

The mists package aims at exploring missing values for temporal data analytically and graphically. It implements a compact abstraction for efficiently indexing missing data in time, along with numerical and visual methods. It also provides new missing data polishing techniques. The Github repository has received 22 stars, but the package is not on CRAN yet. The documentation site is available at https://pkg.earo.me/mists.