• Vlad Fridkin

Dashboarding with R Shiny

R Shiny is a flexible development platform and dashboarding is a common use case. Packages including shinydashboard, shinydashboardPlus and others help 'specialise' R Shiny for dashboarding - however unlike major specialist dashboarding platforms like Power BI and Tableau, the dominant use case is coding over a mouse driven drag-n-drop interface. One of the many benefits of using code is the ability to use version control with git and share easily - here's the github repository with all the code. If you're new to RStudio - follow the steps below to run it on your own computer.

Let’s make it real

We are two intrepid dashboard explorers, Stuart and Vlad, hi there! Stuart built a dashboard using Power BI and Vlad built a dashboard using R Shiny.

Common to both we agreed on the same start and end point.

  • Start point - a publicly available dataset on Kaggle (video game sales)

  • End point - a dashboard page with 2 sliders, 11 KPI, a single table and a single chart

We took two completely different journeys. In this article, I'll go into steps used to create a dashboard with R Shiny, for a comparison with Power BI there are two companion articles:

  • Dashboarding with Power BI vs Shiny

  • Dashboarding with Power BI - steps for creating a Power BI dashboard

Vlad’s journey - R Shiny

The steps I took in building the shiny dashboard included

  • Set up a git repository for version control (used github)

  • Layout the file system for the app (directories and files)

  • Choose R packages

  • Import and process data with reactives

  • Build modules

  • Deploy (used with free account)

While R Shiny is popular for creating dashboards, it can be used to create almost any type of web application. If one had the time and determination, one could even create a version of Power BI or Tableau using R Shiny and R packages - it’s that flexible. Here we’re sticking to comparability, so we’ll limit it.

Set up git

Building an R Shiny app means writing scripts with code and saving these in text files. This works well with git version control. It is easy to compare different versions when the differences are in text. The files are mostly R code files with file extension ‘.R’ like ‘app.R’ which is a special filename recognised by R Shiny as an app file. Whilst one can use just a text editor and R command line, it’s easier to use the popular RStudio IDE (integrated development environment).

RStudio is an end to end solution for R - letting you connect to any data connection, import and export data, create data science work flows, develop web sites and web apps (such as with R Shiny), HTML presentations, write books and scientific articles and publish them to the web or for printing - I used the open-source version.

Layout the file system

In my experience, those new to R Shiny tend to have one or two large files with all the code. More mature coders spread it out into logically named files each with a small amount of code. This modular approach makes it easier to manage the code base, add features and have multiple team members working on the same project at once.

Choose R packages

Shiny is itself an R package - packages extend R’s functionality and are easy to install (a single command install.packages). The community supporting R Shiny is large and increasing every year. Anyone can develop an R package. Quality control for R packages includes standard formats for the files structure, help system, documentation and publishing on CRAN (the standard source for R packages) and most importantly, as R packages are open source anyone can find bugs in the code itself and report them directly to the developer of the package - this means a lot of eyes can look over the code.

The tricky thing is that it takes time to develop an understanding of which packages to use, as there can be many ways to do the same thing. For example, to build a table in R Shiny one can use one of these seven packages: DT, rhandsontable, gt, reactable, flextable, huxtable, kableExtra - and there are more. Each has different functionalities and ways of using it. It’s even possible to use multiple packages in one project and have tables using different packages. For tables, this is due to package developers converting javascript libraries into R packages. Javascript libraries power web applications, so if you see a popular widget on the web then there may already be an R package that makes it accessible to R Shiny - if not then you can try building it yourself.

Here I used the following packages (in bold) for different components of the dashboard:


  • web app layout and reactivity

  • images and titles

  • sliders


  • KPIs


  • the table


  • the chart

I wanted a specific look for the KPI, so I copied the code from an existing function and modified it - this is possible because the packages are open source.

Import and process data with reactives

As with displaying and interacting with tables, there are also multiple packages for importing data. RStudio even has a menu making it easier to import data as it will automatically generate code for you. In building a web app with R Shiny, I like to import data into a reactive. A reactive stores data in the memory of the computer in a way such that when the user makes a change, the reactive responds to the change. In this particular app, I set up a reactive to import the data and another reactive to filter the data when the user moves the sliders. This approach means that the data doesn’t have to be re-imported every time the user makes a change - improving app performance. With more complicated apps, there can be a long trail of reactives, each one responding to something different. This design means that calculations can be minimised to just those necessary - maximising performance. This level of control is critical with larger and more complicated applications.

Build modules

We already discussed that breaking code up into small files has many benefits. R Shiny supports modular development with the ability to manage module name spaces. This means you can name code objects (such as reactives storing data, or UI components such as buttons) the same name in different contexts instead of always coming up with a unique name for each new object. Coders constantly have to come up with names so this is a life saver as the app grows in size and complexity. This app is small, so the benefit of modular development is not significant - however it’s a good habit to have so I used it anyway - and in a work environment, you never know that the small and simple app might tomorrow be required to add more and more functionality.


Luckily the folks at RStudio make it very easy to deploy an app for free on the internet for anyone to visit and use. Just set up an account with and publish with a button click in RStudio. There is a limit of 25 hours per month and your app can be hosted in perpetuity.

The code is available here.

17 views0 comments

Recent Posts

See All
We are DATA, We are Maltem Australia

Maltem Australia acknowledges the Traditional Custodians of Country throughout Australia                                  © 2020 by Maltem Australia

  • LinkedIn