EKtheSage
commented
5 years ago Do we need to include the python's deployment framework? My colleague at work likes to talk about how easy it is to use flask to deploy model. Personal;y I am fine either way, as long as we can integrate our model with the IT team smoothly.
Blog posts about using docker and R to deploy models have been showing up more recently:
- Tmobile TF deployment with plumber: https://github.com/tmobile/r-tensorflow-api
- Telethon Kids Institute docker with HTTPS connection for SSL encryption: https://telethonkids.wordpress.com/2019/03/26/use-rstudio-server-in-a-virtual-environment-with-docker-in-minutes/
- Shinyproxy + RStudio Server: https://www.rkingdc.com/blog/2019/3/27/approximating-rstudio-server-pro-and-shiny-server-pro-and-jupyterhub-for-free
- Docker and R plumber with SSL: https://qunis.de/how-to-make-a-dockerized-plumber-api-secure-with-ssl-and-basic-authentication/
kevinykuo
This is meant to be pretty general but let's take an underwriting model for a personal auto carrier as an example. The starting point is a trained u/w score model and decision rules on whether to accept/deny/defer to human an auto insurance application. We want to expose this model as a service so apps/websites can hit it.
This is impactful/important because we need to demystify the operationalization process, and data scientists (incl. actuaries) should be familiar with the entire stack even if their day-to-day isn't orchestrating containers.
Initial assumptions/requirements
Results