Before Juypter and R Shiny, data scientists needed to become web experts in order to serve results as a web page. This barrier was so high that many reports were copied into Word documents or Powerpoint slides, often manually. With Jupyter and R Shiny, data scientists are now able to communicate results to stakeholders with small applications written only in their language of choice. Reports can be created automatically, which takes less time and improves reproducibility.

Generally, Jupyter and R Shiny are quite user friendly. As the ecosystem is in active development, however, many new libraries are appearing and changing. Best practices are still being refined. In this talk I will share the experience of my team building apps in Python and R at Civis and offer some guidance (with examples!) on our favorite tools. In particular, this talk will touch on

• Client-only and client-server applications: The benefits and drawbacks of each.
• Key components of apps: Widgets and layouts.
• Tools outside of Jupyter and Shiny (Dash, Pyxley).