Smart Finance Budget Tracker Using

Machine Learning

Name ID
ABIR HASAN 21-45672-3
MD. FAISAL 21-45670-3
NAHIYAN AZIZ 21-45811-3
SYEDA FAIZA MAHBUB 21-44432-1

SUPERVISED BY:

VICTOR STANY ROZARIO

Assistant professor, Special Assistant [CS]
DEPARTMENT OF COMPUTER SCIENCE

RESEARCH PROPOSAL SUBMITTED IN THE PARTIAL FULFILMENT OF THE

REQUREMENTS FOR THE DEGREE OF BACHELOR OF SCIENCE

DEPARTMENT OF COMPUTER SCIENCE AND ENGINEERING

FACULTY OF SCIENCE AND TECHNOLOGY

AMERICAN INTERNATIONAL UNIVERSITY – BANGLADESH

Introduction
The rising complexity of personal and institutional financial management, traditional budgeting
tools often fall short in providing predictive insights and personalized recommendations [1].
Budget tracking entails keeping track of and evaluating an individual's or an organization's
earnings and outlays over a specific time frame. Due to the fast-paced nature of our society
today, many people are searching for effective time and money management strategies[2].

To address this challenge, we propose the development of a Smart Finance Budget Tracker
using Machine Learning. This system will intelligently analyse user spending patterns, predict
financial risks, and suggest optimized budget strategies[3]. Unlike conventional tools, our
solution aims to integrate explainable AI (XAI) and utilize real-world data to enhance
transparency, reliability, and decision-making for end-users[4]. One of the most important
aspects of modern living is efficiently managing one's personal finances. Having a trustworthy
tool to help with money management is crucial in a time when financial decisions have a
significant impact on our present and future well-being [5].

The complexity of modern financial landscapes is evident in the proliferation of financial

products and services [6]. However, the strain on the entire digital banking infrastructure has
increased dramatically because of the COVID-19 outbreak and the requirement that people rely
only on digital payment methods during their isolation. After the COVID-19 outbreak, digital
payments have skyrocketed [7]. An additional study assessed the efficacy of an AI-based
financial advisor that comprehended users' financial inquiries and offered tailored suggestions
using natural language processing (NLP). It was discovered that the system's suggestions were
precise and beneficial, which enhanced user happiness and financial security [8]. An expense
tracker is essential for encouraging appropriate financial behavior and accomplishing financial
goals, whether it is used for personal or commercial management. [9]

Research Gaps
Despite various finance-tracking tools and ML-based models emerging in recent years, several critical
limitations remain:

(1) Data Limitations

● Lack of External Factors Integration:

Existing models predominantly rely on internal financial indicators like income,
expense patterns, or company-specific factors [10] such as corporate governance and
managerial decisions.

Gap Identified:

○ Minimal consideration of macroeconomic indicators (e.g., GDP growth,

inflation rates, interest rates) [11].
 ○ Absence of industry-specific risks that directly affect financial behaviour and
creditworthiness.

(2) Lack of Real-World Validation

● Problem:
○ Current models have only been validated at the prototype level with limited
users [12].
○ No benchmark comparison has been done against leading financial tools such
as Mint, YNAB (You Need A Budget), or PocketGuard.
○ Longitudinal impact on users’ financial literacy or behavioural change remains
unexplored [13].

(3) Lack of Explainability

● Challenge:

○ Most deep learning-based financial risk models operate as black-box systems.

[14]
○ There's limited clarity on feature importance, decision rationale, or model
confidence [2].

● Concern:

○ For regulatory and institutional adoption, AI models must provide interpretable

outputs, especially in financial decision-making environments.

Proposed Approach

a. Enhanced Dataset Integration

● Acquire large-scale datasets from open government banking portals, National Bank,
central banks’ APIs.

● Include external economic indicators, user behavioural data (with consent), and
anonymized transactional records [15].

b. Use of Explainable AI (XAI)

● Implement interpretable ML models such as:

○ XGBoost with SHAP (Shapley Additive Explanations) values.

 ○ LIME (Local Interpretable Model-agnostic Explanations) for user-specific
financial predictions [16].
● Ensure each budget recommendation or financial flag includes a transparent
explanation [17].

Conclusion
The proposed Smart Finance Budget Tracker offers an innovative leap beyond current
budgeting platforms by embedding macroeconomic awareness, real-world validations, and
explainable AI. By addressing key research gaps particularly data richness, interpretability, and
long-term behavioural tracking our system has the potential to become a trustworthy and
insightful financial assistant for individuals and institutions alike. With scalability and
transparency at its core, this solution aims to set a new standard in financial AI systems.

Reference
[1] O. Hean, U. Saha, and B. Saha, “Can AI Help with Your Personal Finances?,” Appl. Econ.,
Jan. 2025, doi: 10.1080/00036846.2025.2450384.

[2] A. Saxena, H. Joshi, M. Sunil, and K. Dangi, “Income and Expense Tracker-an Android
Application,” Int. Res. J. Mod. Eng. Technol. Sci., no. 11, pp. 2582–5208, 2020, [Online].
Available: www.irjmets.com

[3] D. Brilliantov and L. Mandrikova, “SMART BUDGETING WITH AI: TRANSFORMATIVE

APPROACHES IN PERSONAL FINANCE MANAGEMENT,” Open Inf. Comput. Integr.
Technol., vol. 0, no. 103, pp. 100–111, May 2025, doi: 10.32620/OIKIT.2025.103.08.

[4] W. Shi, L. Xu, and D. Peng, “Application of Deep Learning in Financial Management
Evaluation,” Sci. Program., vol. 2021, no. 1, p. 2475885, Jan. 2021, doi:
 10.1155/2021/2475885.

[5] Bhimasen N K, Kiran Poorna Vishnu M, Nandeesh B K, and Ninad, “(PDF) Sly Spend: AI
powered personal finance management app.” Accessed: Jul. 07, 2025. [Online]. Available:
https://www.researchgate.net/publication/371219800_Sly_Spend_AI_powered_personal_finan
ce_management_app

[6] Sai Deepak Talasila, “(PDF) AI-Driven Personal Finance Management: Revolutionizing
Budgeting and Financial Planning.” Accessed: Jul. 07, 2025. [Online]. Available:
https://www.researchgate.net/publication/382679575_AI-
Driven_Personal_Finance_Management_Revolutionizing_Budgeting_and_Financial_Planning

[7] V. Kanaparthi, “AI-based Personalization and Trust in Digital Finance,” Jan. 2024, Accessed:
Jul. 07, 2025. [Online]. Available: https://arxiv.org/pdf/2401.15700

[8] N. Wang, Y. Liu, Z. Liu, and X. Huang, “Application of Artificial Intelligence and Big Data in
Modern Financial Management,” Proc. - 2020 Int. Conf. Artif. Intell. Educ. ICAIE 2020, pp.
85–87, Jun. 2020, doi: 10.1109/ICAIE50891.2020.00027.

[9] Neha Jain, Devanshu Mishra, Akash Sahani, and Himanshu Prajapati, “EXPENSE
TRACKER.” Accessed: Jul. 07, 2025. [Online]. Available:
https://www.researchgate.net/publication/388499664_EXPENSE_TRACKER

[10] D. Seretidou, D. Billios, and A. Stavropoulos, “Integrative Analysis of Traditional and Cash
Flow Financial Ratios: Insights from a Systematic Comparative Review,” Risks 2025, Vol. 13,
Page 62, vol. 13, no. 4, p. 62, Mar. 2025, doi: 10.3390/RISKS13040062.

[11] K. Veeraiah, S. V. S. S. S. Raju, Y. Vangaveti, V. V. Kumar, and S. S. Ali, “Role of Artificial

Intelligence in Financial Management,” 2022 Int. Conf. Comput. Commun. Informatics, ICCCI
2022, 2022, doi: 10.1109/ICCCI54379.2022.9740900.

[12] Corina Sas and Mariam Zaid Alenazi, “Evaluating Budgeting Apps: Limited Support for
Budgeting Compared to Tracking.” Accessed: Jul. 07, 2025. [Online]. Available:
https://www.researchgate.net/publication/377242109_Evaluating_Budgeting_Apps_Limited_S
upport_for_Budgeting_Compared_to_Tracking

[13] M. J. D, K. Venkatesh, J. K. K, N. G. R, and V. K. N, “Budget Buddy: An AI-Powered

Finance Tracking Solution for Smarter Money Management,” Ijarcce, vol. 14, no. 3, pp. 496–
504, 2025, doi: 10.17148/ijarcce.2025.14364.

[14] A. Hosoyamada and T. Yamakawa, “Finding Collisions in a Quantum World: Quantum Black-
Box Separation of Collision-Resistance and One-Wayness,” Cryptol. ePrint Arch., 2018,
Accessed: Jul. 07, 2025. [Online]. Available: https://eprint.iacr.org/2018/1066

[15] S. Aishwarya and S. Hemalatha, “Smart Expense Tracking System Using Machine Learning,”
no. AI4IoT 2023, pp. 634–639, 2024, doi: 10.5220/0012613900003739.

[16] L. Pan et al., “Development of prediction models using machine learning algorithms for girls
with suspected central precocious puberty: Retrospective study,” JMIR Med. Informatics, vol.
7, no. 1, Jan. 2019, doi: 10.2196/11728.

[17] Ranu Sewada, Ashwani Jangid, Piyush Kumar, and Neha Mishra, “Explainable Artificial
Intelligence (XAI),” Int. J. food Nutr. Sci., vol. 12, no. 1, Oct. 2023, doi:
10.48047/IJFANS/V12/I1/271.