Designing a Predictive Expense Tracking Mobile Application: A Comparison of Random Forest, Linear Regression, and Prophet-Based Models

Corresponding Author ORCID Identifer

Li Zhe Poh https://orcid.org/0009-0006-3269-7617
Md Ismail Hossen https://orcid.org/0000-0003-3632-937X

Section

Regular Issue

Abstract

In today’s fast-paced world, people engage in various forms of daily consumption, yet many lack the habit of tracking their expenses due to low financial awareness. Without proper expense recording and budgeting, individuals often struggle to understand their consumption habits and are at a disadvantage when it comes to financial planning, especially during emergencies. To address these issues, this project proposes the development of a user-friendly expense tracking mobile application designed to help adults build healthy financial habits and utilize predictive algorithms to forecast future expenses. The application enables users to record daily transactions, analyse their spending patterns, plan budgets, and receive personalized saving recommendations. An experiment was conducted to compare four forecasting models: Random Forest, Linear Regression, Facebook Prophet, and Neural Prophet. The models were evaluated using the same synthetically generated transaction dataset, which yielded a limited number of monthly observations for testing. Random Forest showed the most stable overall performance under this experimental setting and was selected as the forecasting model for the prototype, though these results should be interpreted with caution given the small evaluation sample.

Cite This Article

Tai, L. W., Poh, L. Z., & Hossen, M. I. Designing a Predictive Expense Tracking Mobile Application: A Comparison of Random Forest, Linear Regression, and Prophet-Based Models . Journal of Informatics and Web Engineering, 5(3), 252–265.https://doi.org/10.33093/jiwe.2026.5.3.16