Bayesian network algorithm for predictive modeling of cyber security for efficient bank channels digitalization

Abstract

Bayesian Networks Algorithm (BNA) technique is a vital modelling technique for optimization. The tool is useful for valuable information from the overwhelming volume of data regarding Cyber Security for efficient bank channel digitalization and customer satisfaction. The major threat to electronic banking system are the concerns of security and privacy of information. Basically, banking is highly competitive and is sensitive to economic conditions of many nations of the world. As a result of risks involved, a key strategy for many banks is to improve their performance by reducing costs and increasing revenues. In order to proffer direction for the future work and development, a comprehensive and most up to date review of the current research status of data science model in banking is valuable. The work was carried out in different segment. First, an evaluation on the customer awareness level and adoption of e-banking channels through the procedure of scientific questionnaire. Second, a scientific model for detecting fraudulent transaction given that an instance of a transaction is initiated. In the third phase, the paper adopts the methodology of Data Science Model for the implementation with data mining framework based on Bayesian Net work Algorithm (BNA) . Considering its sophisticated conceptual research framework, which is obviously demonstrated to obtain a transformed data by classification of quantities the paper adopts p(xlNF) and p(xlF) for model training and testing. The outcomes with modified values were compared with real observed results. The residuals were also evaluated in terms of normality, independence and constant variance. The optimized graph indicates a good outcome as the quality and accuracy of the model was evaluated for performance optimization. The experimental results demonstrate the importance of data science model in assessing the impact of cyber security in e-banking services. The implemented model gave 88.5% accuracy on our training sets while test-set model gave 90% accuracy which indicate a better result in tackling cybercrimes in the banking sector. Consequently, by obtaining and carrying out analysis of the trends of research focus, data resources, scientific aids, and data analytical tools, this work contributes to bringing valuable perceptions with regard to the future developments of both Data Science Model and the banking sector along with a full reference summary. Moreover, the paper identifies the key problems and relates the summary for all interested groups that are facing the challenges of Cybers Security threats.