Degree

Bachelor of Science (Computer Science)

Department

Department of Computer Science

School

School of Mathematics and Computer Science (SMCS)

Advisor

Dr. Jawwad Farid, Professor of Practice, Department of Computer Science

Keywords

Deposit Behavior Analytics, Financial Data Engineering, Liquidity Risk Modeling

Abstract

The purpose of this project is to transform raw deposit balance data into meaningful behavioral intelligence that supports decision-making across multiple banking stakeholders, including treasurers, risk managers, branch managers, relationship managers, product managers, and senior management. Traditional banking analysis often relies on static balance snapshots, which fail to capture how deposits behave over time, especially under changing or stressful conditions. To address this, the project introduces a layered analytical framework called the Deposit Behavior Engagement Model, which processes large-scale daily end-of-day deposit data into reusable account-level and segment-level behavioral metrics such as inflows/outflows, volatility, drawdowns, seasonality, persistence, probability, and trigger-based risk analysis. The project contributes by designing a scalable and optimized database architecture capable of handling over 10 million rows of financial time-series data while generating stakeholder-ready analytical reports, on both the frontend and excel, efficiently within practical time constraints.

Tools and Technologies Used

Oracle, SQL, Pro*C, TypeScript, JavaScript, Node.JS, React, Excel

Methodology

The project follows a layered database architecture and analytical pipeline approach designed to transform large-scale raw deposit data into reusable behavioral analytics and stakeholder-ready reports. The methodology begins with importing raw daily end-of-day deposit balances, which are then normalized into a relational schema using mapping and standardization layers to ensure consistency and reduce redundancy. To improve analytical performance and minimize repeated joins, an optimized denormalized account master layer was introduced. On top of this, reusable scratch/base analytical tables were created to precompute account-level and segment-level behavioral metrics such as daily changes, inflows/outflows, volatility, seasonality, and drawdowns. The system further separates parameter independent computations from parameter-dependent analytics to avoid recalculating expensive logic repeatedly. Finally, reporting and view layers expose business-ready outputs for different banking stakeholders. This methodology emphasizes scalability, modularity, performance optimization, and reusability, enabling efficient analysis of millions of rows of financial time-series data.

Document Type

Restricted Access

Submission Type

BSCS Final Year Project

Creative Commons License

Creative Commons Attribution 4.0 International License
This work is licensed under a Creative Commons Attribution 4.0 International License.

Download

Share

COinS