Development of a hybrid system of artificial neural networks and artificial bee colony algorithm for prediction and modeling of customer choice in the market

Abstract

With the increasing growth of technology and the emergence of various industries, numerous manufacturers have entered this field. In today's world, sellers and manufacturers find themselves among a vast number of competitors. Therefore, they need to adopt a variety of policies and strategies for their own survival and profitability. Companies should identify their customers’ needs and adopt their own policies based on customers’ purchase behaviors. To this end, attempts have been made to identify the customer choice model since the past decades. These models aim at modeling and predicting customer choice among several brands. Traditional models were of interest for many years and these methods were frequently used with the advent of artificial intelligence and machine learning systems. They could demonstrate very good results. In this study, it has been attempted to present a new method for the modeling and prediction of customer choice in the market using the combination of artificial intelligence and data mining. Indeed, the new model is to be applied in helping managers with decision-making. Hence, probabilistic neural networks have been combined with artificial bee colony algorithm.  The proposed model was tested in a real market and its efficiency and accuracy were finally compared with those of other models, including neural network trained with back-propagation, probabilistic neural networks, and the neural networks trained with genetic algorithm. The results reveal that the hybrid model shows better performance than the other models.

Keywords: Consumer Choice Model, Data Mining, Artificial Intelligence, modeling, predicting, probabilistic neural network, artificial bee colony algorithm