Bayesian Statistical Analysis On Consumer Purchase Decisions

The rapid progress of science and technology has led to intense market competition,and consumer preferences and demand become diversified.In order to meet the needs of target market more effectively,enterprises need to analyze the consumer’s purchase decision behavior to understand the demand of consumers in target market,do market segmentation more effectively,and position target market more accurately.To solve the problems of consumers purchase decision,we analyse the problems of overdispersion,individual-level parameter estimation and small sample,at the perspective of pre-purchase decision and post-purchase satisfaction.In this paper,using the bayesian theory and method,we study consumer purchase decision on theory and application in group-level parameter estimation and individual-level parameter estimation of pre-purchase decision,and customer satisfaction of post-purchase.We have mainly done the following research in theory:first,using the theory advantage of bayesian,we can effectively solve the problems of the data acquisition difficulties or overdispersion,effectively optimize the traditional theoretical model by Bayesian logit model analysis in group-level parameter estimation of pre-purchase decision.Second,we construct the hierarchical Bayesian random effects model for the problems of individual-level parameter estimation in the actual consumer purchase decisions,effectively solve the problem of the lack of individual consumer data,avoiding the error of traditional research methods,such as least squares estimate.At the same time,in the process of modeling,we use a continuous population distribution to describe the preference difference between individual consumers,and evaluate the uncertainty in the study of consumer preference behavior.Third,in the condition of small samples,we use Bayesian method to study the influence factors of customer satisfaction by the construction of a structural equation model,and obtain the final score of customer satisfaction based on multi-grade score by Bayesian estimation.Fourth,we introduce the Bayesian method and hierarchical Bayesian methods in the study of consumer purchase decision,and make more marketing researchers and practitioners realize the unique advantages of Bayesian method.At the same time,Bayesian theory is applied to the actual consumer purchase decision through the combination of theory and practice,and can develop in the field of consumer purchase decision.In applied research,we make an empirical analysis of the consumers purchase data using the Bayesian method and hierarchical Bayesian model,and effectively solve the problems of marketing strategy,such as overdispersion data,individual-level parameter estimation and small sample,and improve research methods of consumer purchase decision.In the study of group-level parameter estimation in pre-purchase decision,we guide the coffee cup company to carry out marketing activities by 4C marketing mix(consumer strategy,cost strategy,convenience strategy and communication strategy).In the study of group-level and individual-level parameter estimation in pre-purchase decision,through the hierarchical Bayesian random effects model,we not only get the part-worths of yoghurt each attribute and demographics variables influence on part-worths,but also obtain yogurt part-worths of individual consumers.Thus yogurt company can reasonably divide the market,acquire the advantage of the market by improving the yogurt attribute combination,and position the market accurately.In the study of customer satisfaction in post-purchase decision,we give the way to improve customer satisfaction of meters/bonwe,and obtain final score of customer satisfaction by Bayesian estimation.We introduce Bayesian statistical analysis in the consumer purchase decision,do empirical research.This study has the following innovative:first,in consumer purchase decision,the traditional method is uncertain for the problems of heterogeneity parameters and sample quantity limitation.So we build hierarchical bayesian random effect model in consumer pre-purchase decision,not only get the average part-worths estimate,the coefficient matrix and covariance matrix,of all respondents,but also obtain individual consumer part-worths estimate,so that we can effectively position target market.Second,in the face of small sample,we construct the structural equation model of leisure clothing,and estimate small sample data using the Bayesian method.Again because the covariance structure equation model cannot estimate the final score of customer satisfaction in the consumer post-purchase,we use Bayesian method based on multi-grade score,and respectively use three prior distribution to calculate the Bayesian estimate of customer satisfaction.Third,because overdispersion data is easy to underestimate the standard error of parameter estimate and produce inaccurate significance test,we construct the Bayesian logit model,enrich consumer purchase decision system by Bayesian estimate,evaluation and prediction,effectively solve the overdispersion problems exist in the consumer purchase decision,and provide theoretical support and practical advice in marketing strategy.