Summary

The web content provides a comprehensive guide to creating a live interactive dashboard in R Shiny, detailing data sources, transformation, app structure, visualizations, and deployment on a web server.

Abstract

The article presents a step-by-step tutorial on building a dashboard in R Shiny, aimed at analyzing graduate employment and income outcomes in Singapore. It covers the use of two datasets for employment metrics, the process of data cleaning and transformation, and the structure of a Shiny App with a front-end (ui.R) and back-end (server.R). The guide also demonstrates how to create interactive visualizations using Plotly and Kable, and concludes with instructions for deploying the dashboard on the shinyapps.io server. The project's code is made available on GitHub, and the author encourages readers to use and modify it for their purposes, while also expressing gratitude to mentors and collaborators.

Opinions

The author advocates for R and Shiny App due to their free, scalable, and user-friendly nature, supported by a generous community.
The dashboard is intended to aid young individuals in Singapore in making informed decisions about education and employment prospects.
The author finds Plotly and Kable to be sophisticated tools for creating interactive visualizations and tables, enhancing the user experience of the dashboard.
The article promotes the use of shinyapps.io for hosting Shiny Apps, highlighting its ease of use and the fact that it offers free hosting for up to three applications.
The author expresses appreciation for the guidance received from Dr. Piotr Wójcik and Mgr. Piotr Ćwiakowski, as well as the collaboration with Zimin Luo during the project's development.
A recommendation is made for ZAI.chat, an AI service alternative to ChatGPT Plus, emphasizing its cost-effectiveness.

A Complete Dashboard Project in R Shiny App

A step-by-step guide to building a live interactive dashboard in R Shiny. Full codes accessible.

To get a taste of what we will build in this guide you can take a look at the live dashboard: https://gesurvey.shinyapps.io/Graduate-Employment-Survey/

Now, one can ask why R and Shiny App? I would answer - because it is free, scalable, and easy to learn, supported by a generous & enthusiastic community, and it's pretty.

The full codes of the dashboard project on Github.

Content

Data sources
Data transformation
Shiny App structure
Visualizations
Deployment of the dashboard on https://www.shinyapps.io/ web server.

Our motivation is that the dashboard will help young people of Singapore and those seeking study opportunities in Singapore to analyze which university & degree has a better outcome with regards to employment & income.

Let’s get started!

1. Data Sources

We will use two datasets in this project.

Graduate employment survey dataset can be found here: data.gov.sg/ges

The number of graduates by university dataset can be found here: data.gov.sg/gra

Data table about the number of graduates by university

As mentioned the main metrics of the analysis are:

Employment Rate
Monthly Gross Income

We will use the following R packages:

The logos are taken from the library documentation and put together by the author

2. Data transformation

The standard workflow about data transformation looks like this:

In the codes provided on GitHub (DataWrangling.R), we will basically attach dataset, clean NAs, fill missing values, change university names to make it more intuitive, and finally, we will save the data frames in .rds format which is much faster to read at the next steps.

If you see the names of the school unique(data_e$school) in the dataset there are some names that need cleaning, for example:

Transforming factor variables into numeric:

data_e <- data_e %>% modify_at(c(5:12), as.numeric)

And finally, saving cleaned data frame as .rds extension for Shiny:

saveRDS(data_e, file = “data/employment_data.rds”)

3. Shiny App structure

To run a Shiny App you need to install the package & import the library:

install.packages("Shiny")
library(Shiny)

Shiny applications have two important components, I call them front-end ui.R and back-end server.R.

In ui.R we create a structure of front-end, how we want our web application to look like.

This is the basic diagram of the ui for our dashboard:

Diagram of ui (front-end) of our dashboard

Now, let’s run the first example of ui.R and its corresponding server.R codes.

We have front-ent, now we need the corresponding back-end. If you look at lines 21 and 21 we have table output “datahead” and plot output “piePlot”. Let’s configure the server.R.

After running the ui.R or server.R code the result is below:

The full codes are on GitHub for ui.R and server.R. You will need to have ui.R and server.R in the same folder in separate R files. When you run one of the R files the Shiny App will open.

4. Visualizations

Visualizations are mostly done in Plotly and the data tables are presented in Kable. I recommend checking some examples on the internet because they are so fancy! Here are useful links for Plotly interactive graphs and Kable awesome tables.

From the project, I will post here some examples.

Plotly

Let’s dive deeper into the details of the National University of Singapore and list the basic monthly median income for each school. Code example:

Result:

The monthly median salary of graduates by Schools of the National University of Singapore

And we can see from the plot that graduates from the College of Law earn the most :).

Kable

To show the pretty and interactive table of Kable, Let’s see the full table of Fulltime Employment and Balance Monthly Salary (Median) by Universities, their Schools, and Degrees in 2018. Code example:

And the results:

5. Deployment of the dashboard on https://www.shinyapps.io/ web server.

To publish your Shiny App online, you should sign up on www.shinyapps.io, it is free to run three applications.

Please read this guide as it explains in detail what you need to do to create a connection between your app and shinyapps.io server.

Few steps are here:

0. I assume you have ui.R and server.R in the same folder.

Run code in your R file:

install.packages('rsconnect')
library(rsconnect)

2. Create a shinyapps.io account. Please note that shinyapps.io will use your account name as the domain name for all your apps.

3. When you log in to shinyapps.io, get the token that is generated on the website:

https://shiny.rstudio.com/articles/shinyapps.html

4. Configure rsconnect package to use your account. Citing the website: “Click the show button on the token page. A window will pop up that shows the full command to configure your account using the appropriate parameters for the rsconnect::setAccountInfo function. Copy this command to your clipboard, and then paste it into the command line of RStudio and click enter.”

5. Publish your app. To deploy your application, use the below codes:

library(rsconnect)
deployApp()

Then click the Publish button in R (from ui.R or server.R).

Once the deployment finishes voalaaaa!!! Congratulations! Your first Shiny App is published. If you make changes in ui.R or server.R files you can republish the app.

In this step-by-step guide, I showed you how to build this web dashboard in a Shiny App: https://gesurvey.shinyapps.io/Graduate-Employment-Survey/. We reviewed Data sources, Data transformation, Shiny App structure, Visualizations used in the project, and deployment steps. Feel free to access and use the codes. Downloading and running them should give you the same results. Full codes are here. If you run into issues feel free to drop me a message ;)

Special thanks to Dr. Piotr Wójcik and Mgr. Piotr Ćwiakowski — our teachers in Advanced R at the University of Warsaw. And to my colleague Zimin Luo with whom I was working on this project during Graduate Studies in Data Science at the University of Warsaw.