Chair: Richard Payne (Lilly)
Instructor: Will Landau (Lilly)

This hands-on interactive workshop consists mostly of guided R programming exercises in online notebooks.
Each participant must

1. Have basic R programming skills, and
2. Bring a laptop with a web browser and the ability to connect to the internet.

Complicated workflows in pharmaceutical statistics , such as Bayesian network meta analyses, machine learning, and target identification, can be difficult to develop and maintain. When a single round of computation lasts multiple hours and produces several artifacts, it becomes cumbersome and expensive to repeatedly refresh results in response to changing code and data. For the R and Statistics communities, who are accustomed to short runtimes and interactive workflows, the standard workflow management techniques often break down.

This interactive short course offers guided hands-on practice implementing robust data analysis pipelines with the {targets} R package. The exercises begin with the building blocks of a machine learning project and gradually construct a full end-to-end pipeline. Participants experience the benefits of pipelines first-hand: when the upstream code or data changes, {targets} automatically determines which machine learning models need to rerun in order to stay up to date, and it automatically orchestrates those models. This behavior keeps the results synchronized with their dependencies while avoiding wasting time on unnecessary models that could take hours to rerun.

Will Landau, PhD
Eli Lilly

Will Landau is a Research Scientist at Eli Lilly and Company, where he develops methods and tools for clinical statisticians, and he is the creator and maintainer of the targets R package. Will earned his PhD in Statistics at Iowa State University in 2016, where his dissertation research applied Bayesian methods, hierarchical models, and GPU computing to the analysis of RNA-seq data.