Expenses tracker website with AI enhancement

Abstract

This thesis presents the design and implementation of a personal finance tracker that analyzes user-uploaded bank statement files in CSV or Excel format. The system automatically detects file columns, parses transactions, categorizes spending using a combination of keyword rules and a local machine learning model, and stores confidence scores for automated labels. The application generates spending reports with charts, category breakdowns, recurring payment identification, and budget profile feedback. Additional features include monthly trend analytics and a review queue for low-confidence or unlabeled transactions. The final result is a Docker-based modular monolith that prioritizes deployability and practical financial insight without relying on external banking integrations or cloud AI APIs.