Evaluation of the relationship between brand measures and customer satisfaction by using data mining techniques

Different measures have been proposed to study brands. In this paper, it is studied whether regression methods can capture the relationship between customer satisfaction and brand measures. It is also investigated whether a combination of these brand measures is useful for the prediction of customer satisfaction. Various regression methods were employed and it was found that generally there was a high correlation (>0.7) between the combination of brand measures and customer satisfaction. Attribute selection methods were used to find out the most important components among all the components of different brand measures. Results suggest that a small subset of all the components (7 out of 111) gives almost the same prediction accuracy as with all the components of different brand measures. This subset of components consists of components from different brand measures. The results emphasize that various brand measures should be combined to improve the prediction accuracy of customer satisfaction. Experiments also suggest that while various regression methods produce good results, support vector machine regression method generally perform best for this problem.