Research on the Influential Factors of Customer Satisfaction and Post- Purchase Behavior for Hotels: The Multilayer Perceptrons Neural Network Approach and Logistic Regression Analysis

The majority of previous studies have used logistic regression analysis(LRA) as an analysis tool to evaluate the impact of service attributes on customer satisfaction and post-purchase behavior. However, LRA assumes that a linear relation exists between the input and output variables, and that variables are normally distributed, which places significant limitations on the analysis. This study uses multilayer perceptrons neural networks(MLPs) to overcome the limitations of LRA. The results show that among the four hotel service attributes, “personnel services” has the greatest impact on customer satisfaction and repurchase or recommending the service to others, whereas “business and travel services” has the lowest impact. MLPs is more accurate than LRA. MLPs achieved an accuracy rate of 93% for predicting customer satisfaction, and 90% for predicting repurchase or recommending the service to others. Although LRA has an accuracy rate of 87% for predicting customer satisfaction, it only scores 23.77% for predicting “unsatisfied customers.” MLPs has an accuracy rate of 69.23% for predicting this category. LRA has an accuracy rate of 80% for predicting repurchase or recommending the service to others, but only has 9.52% accuracy in the “no intention to repurchase or recommend the service to others” category. By contrast, MLPs has an accuracy of 71.43% for this category, indicating that MLPs can predict repurchase or recommending the service to others more effectively.