Advent of “Grammar”: Bridging Statistics and Data Science for the Design of Experiments

Experimental Design
Author
Affiliation

Monash University

Published

November 27, 2020

Abstract

Statistics is a valuable tool for almost all scientific fields and industry to make sense of their data, yet as a field we lag behind to remain relevant, getting superseded by the so-called data science. What differentiates statistics from data science? And how does the design of experiments fit in with data science?

In this webinar, I will talk about the concept of the “grammar” and the momentum it is gaining to make data analysis more accessible to a diverse group. I’ll then present my prototype to the “grammar of experimental design” - a framework to construct the design of comparative experiments that cognitively enforces the experimental structure. I’ll explain some principles behind this prototype, showcase how my developmental R-package edibble will work, and how I think it helps to bridge the gap between experimental design theory and practice for the wider community.

Errata

There was an error in my script (where duplicates were not removed) for the count for the number of R packages in CRAN Task View of Design of Experiments. The slides specify there are about 200 R-packages. After removing duplicates, there was about 100 R-packages.

Click here for the link to the slide.