This is the source code I developed to showcase my dexterity with Data Science and Data Engineering practices, such as clustering, production deployment of web applications, and friendly GUIs. In fact, I deployed it as web application on Heroku. A working alternative is to deploy it on AWS EC2, even though - for this small-scale project - costs would be higher and improvement in performance unimpressive.
The structure of the present repository follows best practices as long as the modules used allow bug-free code (e.g. dash).
This is the main dash application running on Flask. It also includes callbacks for the dynamic features of the web application.
It includes settings to be changed depending on whether you are working locally or deploying in production, by simply changing the value of the variable ENV. It automatically sets the debug variable to false once in production, due to security concerns. Worth to be noted, this script also includes the code snippet for the PostgreSQL engine, if preferred as alternative.
Minimum versions of the libraries used, necessary for deployment of the application.
Necessary settings for serving the Flask application.
Scripts in this folder are organized by customers' feature. Each script generate the graphs - and relative variables - that are shown once app.py is launched.
It also contains the processed dataset with all the customers' records.
Code efficiency could be improved by refactoring how variables are generated.
📃 favicon.ico: Personal logo shown in the browser tab.
📃 app_layout.css: CSS style of the webapp.
📃 graph_layouts.py: Layout of the tables, generated then in data folder. Similar technique could be implemented for other charts, allowing further shrinkage of the code length.
The preceding clustering technique has been further explained in a separate notebook (TBD).
For an overview of projects in my Portfolio, please follow this link.