An Artificial Intelligence-Based Framework for Intelligent Fund Accounting Transformation

Abstract

The investment fund sector generates huge amounts of financial transactions, which need proper accounting, reconciliation, and management of regulations. Conventional fund accounting systems tend to rely on human intervention, which poses difficulties of operational efficiency, error- reduction, and scalability. The study suggests a framework with Artificial Intelligence to convert fund accounting functions using machine learning to detect anomalies. Since, relatively few datasets, which cover the public fund accounting, are available, a proposed synthetic dataset, consisting of 10,000 financial transaction entries, was created in order to approximate investment accounting processes. Four machine learning models are deployed and compared on the basis of accuracy, precision, recall, F1-score, ROC-AUC metrics to evaluate Logistic Regression, Random Forest, XGBoost and Support Vector Machine with around 97% accuracy. The findings indicate that AI-based solutions can be used to successfully detect irregularities in accounting and enhance operational intelligence. The study provides information that adds a research-backed model that could aid in intelligent automation of the future, monitoring risks and decision-making in investment fund management.