Built as my Capstone project, this webapp helps people working in the B2B Finance working environment to create and manage their invoices. Users would be able to:
- View the invoice data from various buyers
- See various attributes of the invoices from a particular buyer
- Perform Data Pre-processing on the invoice data.
- Get account-level analytics to easily visualize and interpret data
- Get a prediction of when the invoice is going to get paid
A fully responsive webapp built using a ReactJS frontend, Java backend, MySQL database, fueled by XGBoost regression and Flask API:
- Python
- Flask
- JavaScript
- ReactJS
- MUI 5.0
- Java
- JDBC
- Servlets
- MySQL
- A dense tabular dataset of 50k records was pre-processed
- Categorical features were label encoded and output values were binned
- XGBoost was chosen after comparing multiple models, later tuned by Optuna and the model was saved
- A responsive Accounts Dashboard was built, which visualized data as grids and graphs
- CRUD operations, soft refresh and an extensive Search was also added for better UX
- User gets the predicted binned value at a click of a button
- A MySQL database was used for data loading and CRUD functionality
- Flask API was the python backend to communicate with the frontend and return results from our ML model
- Java, JDBC and Servlets were used to receive the GET/POST requests and communicate with MySQL
