Shiny Revolutionized Our Business

Introduction

• Adi Sarid

  • PhD in Operations Research

  • CEO of Sarid Research Institute

  • Working in Data Science for Market Research and Consumer Insights

  • An R-user since ~2007

• In the next 15 minutes, I will talk about how Shiny transformed our business, how we use it, and how you can too.

What is Shiny?

• Shiny is a web application framework for R

• You use a package (shiny) and the familiar R syntax

• Shiny turns your code into an interactive web app

• Benefits

  • The application is online (others can use it as a cloud service)

  • The application server has all the power of R

  • Your code stays secret

  • Doesn’t require HTML / JavaScript (almost)

Example

We created ACME Coffee as a demonstration of some of shiny’s capabilities.

Our Business Before and After Shiny

• Before - data changes, but reports are static

  • Most of our deliveries were static PowerPoint presentations or Word documents

  • Simplistic dashboards based on 3rd party solutions

• After - more projects, revenue increase

  • Deliveries include dashboards - allow us to provide ongoing data solutions

  • Pricing models are more diverse (cloud services)

  • Ability to handle complex projects (optimization, model explainability)

When to Shine?

• Use shiny if…

  • Real-time required (e.g., pulling from database)

  • Application needs user input to filter/analyze dynamically

  • Complex visualizations that benefit from user manipulation

When to Shine? Examples

• MOAG Climate influence dashboard

• Model explainability (e.g., predicting no-show)

• CSAT Survey dashboard

• Teaching, e.g.: food consumption, population vs. sample

• Experiments, e.g.: interactive learning

When NOT to Shine?

• Don’t use it when…

  • Content can be conveyed through static reports/dashboard (e.g., flexdashboard)

  • One-off analysis not requiring interactivity

  • No time for development or limited deployment options

Tech Stack

Shiny can be extremely flexible, depending on the type of deployment. Our stack includes:

• Deployment on AWS EC2 with nginx OR deployment on shinyapps.io

• AWS RDS MySQL for database support

• AWS S3 for storage of “stuff”

• AWS Cognito or customized solutions for user-secure login, if needed

Takeaway

• An empowering tool that turns data scientists into web app developers

• Short learning curve, huge benefit

• Know when to use (and when not to)

Some Recommended Sources

• Mastering Shiny by Hadley Wickham

• Talk by Joe Chang about the inception of Shiny (“The Past and Future of Shiny”) from posit::conf 2022

• Shiny Gallery

Reach out

• Adi Sarid

• adi@sarid-ins.com

• https://www.sarid-ins.com

• I’m writing a Hebrew guide book for R, freely available here:

  • The Hebrew guide to R hebrewr.co.il