A Smart Comparative Analysis For Secure Electronic Websites

Online banking is an ideal method for conducting financial transactions such as e-commerce, e-banking, and e-payments. The growing popularity of online payment services and payroll systems, however, has opened new pathways for hackers to steal consumers' information and money, a risk which poses significant danger to the users of e-commerce and e-banking websites. This study uses the selection method of the entire e-commerce and e-banking website dataset (Chi-Squared, Gini index, and main learning algorithm). The results of the analysis suggest the identification and comparison of machine learning and deep learning algorithm performance on binary category labels (legal, fraudulent) between similar datasets, and understanding which function plays a vital role in predicting safe e-banking and e-commerce website datasets. The e-commerce and e-banking website dataset was compiled from the UCI machine learning library. We obtained 11,056 entries based on 30 unique website attributes. We used the machine learning algorithms support vector machine (SVM), k-nearest neighbors, random forest (RF), decision tree (DT), and the multilayer perceptron (MLP) deep learning algorithm to analyze the datasets of e-commerce and e-banking websites and found the best algorithms based on accuracy, precision, recall, and F1-measure. MLP had the highest precision at 97%. With this procedure we can now accurately test websites to assist in the early prediction of secure e-banking e-commerce transactions.