Publications & Journal Articles

Applying Supervised Machine Learning Algorithms for Fraud Detection in Anti-Money Laundering


As international money transfers become more automated, it becomes easier for criminals to transfer money across borders in a fraction of a second, while it also becomes easier for regulators to inspect and monitor international money mobility and identify unusual patterns of money movement. Machine learning algorithms may be a useful addition to the current money laundering detection issues. This research empirically tested four machine learning algorithms (Logistic regression, SVM, Random Forest, and ANN) using a synthetic dataset that closely matches regular transaction behavior. After observing the performance of different algorithms, it can be stated that the Random Forest technique, when compared to the other techniques, provides the best accuracy. The least accurate approach was the Artificial Neural Network (ANN).

Macro-Economic and Bank-Specific Determinants of Credit Risk in Commercial Banks


 (2021): Empirical Quests for Management Essences IT

Inadequate credit risk assessment procedures may have a significant negative influence on a financial institution's operational performance, perhaps leading to liquidity concerns. It is hypothesized that different factors such as macroeconomic, and bank-specific factors affect the credit risk in financial institutions. The objective of this study is to check those factors responsible for credit risk. The data came from WDI and Bankscope databases. The data is balanced panel data of 106 private and state-owned commercial banks for 6 years (n=106, t=6). This study used Fixed Effect (FE), and Random Effect (RE) models. The results suggest that if inflation, interest rate, unemployment increase, the credit risk of commercial banks increases. The results also suggest that if GDP growth, efficiency, and bank size increase, the credit risk become minimized. Additionally, the credit risk is lower in private banks than in state-owned banks. The findings of this research, however, do not support the hypotheses that exchange rate and regulatory capitals influence credit risk.

Segmentation of Bank Consumers for Artificial Intelligence Marketing


International Journal of Contemporary Financial Issues

Banks can offer more personalized products and services by using segmentation solutions. By gaining a deeper understanding of client characteristics, marketers can Choose the appropriate promotional content to deliver, choose the right marketing channels for the target market, identify new and profitable market sectors, and introduce new products and services. Artificial intelligence marketing uses artificial intelligence concepts and models such as machine learning and Bayesian networks. Cluster analysis is a machine learning method for classifying entities into groups that are homogenous in terms of observable characteristics. This study included K-means cluster analysis, Elbow, and silhouette approaches to segment the data for cardholders of various banks. According to the results from Elbow and the silhouette, the ideal number of clusters seems to be five. Based on their income and shopping frequencies, which are supposedly the greatest attributes to establish the segments of the customers, this research identified five distinct consumer segments: Savers, General, Targets, and Big spenders. This research recommends leveraging machine learning techniques to devise various marketing tactics and policies to maximize the bank’s efficiency, customer satisfaction, and quality of services.