Consumer Behavior and Satisfaction Rate in Social Media Purchasing Ads Using Radial Basis Function Networks (RBFNS) Based BAT Algorithm

Companies who wish to enhance their marketing plans must first be quite conscious of consumer behaviour and satisfaction with social media campaigns allowing purchase capability. Many unstructured data generated by social media channels makes it challenging to find feasible application concepts. Usually, the conventional approaches do not sufficiently show the non-linear interactions between consumer reactions and advertisement effectiveness. While current machine learning methods such artificial neural networks (ANN) or Support Vector machines (SVM) can offer insights, they lack the optimising force required to properly fine-tune parameters. The forecasts of satisfaction rates thus differ from ideal. Better predictive accuracy is obtained in the technique by using a Bat Algorithm in concert with Radial Basis Function Networks (RBFNs). The Bat Algorithm maximises RBFN parameters, which include weights, spread, and centers, so improving their performance. Ten thousand consumer records from social media buying campaigns made up the data used for the study. Among the data were click-through rates and degrees of satisfaction. For model validation as well as training, an 80–20 split was used. Predicting consumer satisfaction with a 96.2% accuracy, the RBFN-Bat model outperformed already in use approaches including artificial neural networks (92.5%) and support vector machines (89.7%). Moreover, the mean absolute error (MAE) dropped to 0.045 and the computational efficiency rose by 18% in comparison to earlier applied techniques.