FINTRACK: An Intelligent Expense Tracking and Behavioural Analysis System

Personal expense tracking applications commonly record transactions but provide limited support for interpreting spending behaviour over time. This paper presents FinTrack, a locally deployable web-based system that combines expense recording, machine-learning-based behavioural classification, trend visualisation, rule-based recommendations, and month-to-month drift alerts. The application is implemented using Python and Streamlit, with SQLite for persistence, Pandas and NumPy for data processing, Scikit-learn for classification, and Matplotlib for visualisation. A Decision Tree classifier uses three features—expense-to-income ratio, savings rate, and number of active spending categories—to assign one of three labels: Saver, Balanced, or Overspender. The model is trained on 360 synthetically generated labelled profiles and evaluated using a stratified 80:20 holdout split. The project report records an accuracy of 97.22% (70 correct predictions from 72 test samples). The system also applies threshold-based alerts for increases in expenditure, declines in savings, and category-level spikes, and provides a dashboard for examining monthly and category-wise patterns. These results demonstrate the technical feasibility of integrating lightweight, interpretable analytics into a personal finance application. However, because the classifier is evaluated on synthetic data generated from rule-based class boundaries, the reported accuracy should not be interpreted as evidence of generalisation to real-world users. Further evaluation on representative real transaction histories is required.