AN INVESTIGATION TO DISCUSS THE MODELLING OF MASSIVE CROSS EFFECTS IN PURCHASE INCIDENCE: COMPARING ARTIFICIAL NEURAL NETWORK TECHNIQUES AND MULTIVARIATE PROBIT MODELLING.

Abstract

Consumer behaviour across related product categories may be better understood by the modelling of huge cross-effects in purchase incidence, which is the focus of this work. In order to capture complicated cross-category interactions, it evaluates the performance of Artificial Neural Networks (ANNs) and Multivariate Probit (MVP) modelling methodologies. When it comes to big datasets and nonlinear interactions, ANNs are recognised for their flexibility, whereas MVP models provide a defined statistical framework based on economic theory. Research assesses these approaches by looking at how well they forecast outcomes, how easy they are to understand, and how well they reflect the dynamics of purchases across different categories. While ANNs do quite well in terms of prediction accuracy, the results show that MVP models provide far better interpretability and theoretical consistency. Practical insights for marketing strategies and decision-making are offered by the results, which also contribute to expanding approaches in consumer choice modelling. By contrasting and comparing the efficacy of Artificial Neural Network (ANN) methods with Multivariate Probit (MVP) modelling, this research investigates large-scale cross effects on sales incidence. This study aims to compare and evaluate two state-of-the-art approaches with the idea of revealing how well they understand complex linkages and can predict consumer behaviour in massive datasets. This research aims to evaluate the methodology's ability to deliver more accurate and practical insights into consumer behaviour by using ANN and MVP algorithms to purchase incidence data. Focussing on model correctness, computation efficiency, and interpretability, this comparison study may help researchers performing marketing or consumer research choose the right modelling methodologies.