Thursday 3rd December 2020, 11:10–11:25 (Australia/Sydney), Zoom Breakout Room 1
Producing output tables is an exceedingly manual activity, particularly when tabling complex statistics with associated metadata. When preparing large numbers of tables for presentation or publication, providing different views of the same result set can require large amounts of re-processing and fiddly manual combination and reshaping of outputs.
Inspired by the gt R package (https://gt.rstudio.com/), we present an experimental proof of concept for a new tabling approach with an accompanying R package, projectable.
Designed to easily support flexible specification and table manipulation, this approach treats a table as a collection of calculations with accompanying metadata rather than simple values. It aims to support easily moving from specification to data production to presentation of results by treating an output table as a “projection” of complex results.
This presentation introduces the architecture and approach behind the package and provides some simple practical examples of real-world use cases.
Kinto is a data scientist at the Social Research Centre. Nowadays he works primarily in R, but has previously worked with Python and Javascript. One could say he came to data science via an unconventional path, having studied physics and philosophy at university, but then again everyone in data science seems to have arrived there in a strange way.
He holds degrees from the University of Melbourne and Monash University, and received numerous awards for academic excellence during his studies, including the Hastie Scholarship and the Peter J. Lloyd Prize in Theoretical Physics.