Research on Factors Affecting Consumer Purchasing Behavior on E-commerce Website During COVID-19 Pandemic based on RBF-SVM Network

The proposed model is able to forecast online purchases by creating a connection between site visitors’ behaviors and the information to which they are exposed. Many modeling challenges arise when attempting to predict purchases made on the Internet. Models of online purchase need to consider the heterogeneity of users because online retailers reach a wide variety of people in a variety of competitive settings. The low probability of making a purchase online also reduces the predictive and explanatory power of models. Customers must complete the product configuration, fill up their personal information, and confirm the order with their payment information before making a purchase. The proposed strategy requires the processing of data, the selection of features, and the training of a model. Natural language processing and normalization are examples of preprocessing that can be used to data collected from e-commerce websites. Specifically, the Particle Swarm Optimization (PSO) technique is employed to select the features. Models are trained using an RBF-SVM hybrid approach. The proposed methodology outperforms SVM and RBF, two popular current methods.