#145 Deploying ML Models Trained in R as APIs

Using the renv package, Joe Brillantes will demonstrate how to recover a Docker-containerized Jupyter development environment with the same versions of R packages previously used. He will then demonstrate how to deploy an R machine learning model in a Docker-containerized web API using the plumber package. This is the continuation of meetup #144 (https://youtu.be/wWIceo2LcUo).

Isolating environments is about eliminating the "it works on my machine" problem. In a traditional setup, your project depends on the specific versions of languages, and system libraries installed on your laptop. If a teammate has different versions from your laptop, the code might break for them. Docker fixes this by wrapping the entire environment into a portable image. "The renv package helps [developers] create reproducible environments for [their] R projects" (https://rstudio.github.io/renv/articles/renv.html). "Plumber allows developers to create a web API by merely decorating [their] existing R source code with roxygen2-like comments" (https://www.rplumber.io/).

Joe is a co-organizer of the R User Group – Philippines. He is currently a data insights and AI specialist in the transnational dairy cooperative FrieslandCampina (Alaska Milk Corporation). He has been working as an advanced analytics specialist since 2008, and has been leading analytics teams since 2015. He has a Master of Mathematics (Business Track) degree from Ateneo de Manila University, and a Bachelor of Science degree in Mathematics Education from Brigham Young University - Hawaii.

Click here to join the meeting: https://teams.live.com/meet/9328890818261?p=S8icJdXQeXhL8RfA8L
Meeting ID: 932 889 081 826 1
Passcode: Fz9fH3