Brazillian e-commerce dashboard 📦

The following dashboard (based on an aggregated version of the e-commerce dataset by Olist ) has been built to explore a few shiny core concepts:

• Reactivity: data has been preaggreated in order to take heavy computaiton away from the user, providing a faster and more reactive user experience.
• shinydashboard structure for faster dashboard styling and layout. Additional UI elements (like the right sidebar) are provided by shinydashboardplus package.
• User authenthication and password hasing is performed through shinyauthr. Usernames and passwords are stored in a MySQL database.
  This authenthication method is not meant to be used in a production environment as it requires to implement logic to check for client authenthication on every piece of content.
• In order to anticipate multi-user usage, plots are cached via redis, halvening load times to render cached plots.
• The app (and the mock user db) has been dockerized for easy deployment.

Overview

The dashboard conists in two views:

1. Sales

   • Order volumes plot over time.
   • Sales brekdwon by state / city.

2. Operations

   • Order approval time by state/city.
   • Delivery processing time by state/city.
   • Shipping time by state/city

The right sidebar allows to:

• Filter by relevant dimensions.
• Change reporting window (daily, weekly, monthly).
• Download filtered data in .csv format.

Usage

Installation

docker and docker-compose will be neede to build and run the containers.

To download and build the app just run:

# Clone the repo
git clone https://github.com/andodet/shiny-test

# Build the image
cd shiny-test
docker-compose up --build

This will likely take a while as a number of R packages will be installed with all their dependencies. After the process is finished the dashboard will be exposed on http://localhost:3838/app

Further development

This dashboard is meant to be a proof of concept and it would still require a few tweaks before being ready for deployment. Namely, it's needed to:

• Setup proper authehtication (via shinyproxy or behind and nginx server).
• Unit testing.
• Use secrets for db credentials.
• Set up a password for redis instance.