【作者】 章明珠；

【导师】 李纯青； 李刚；

【作者基本信息】 西安工业大学 ， 管理科学与工程， 2011， 硕士

【摘要】 非契约环境下,客户未来购买行为逐渐成为营销界研究的热点,学术界通常用随机模型对客户的未来购买行为进行预测,但这种模型在实践界的使用相对较难。如何探索一种简单实用的方法是本研究的主要任务。本文以某商业银行活期储蓄客户的历史交易数据为研究样本,构建启发式算法和随机模型,对客户未来购买行为进行预测,从实证角度证明了启发式算法的确比随机模型预测方法更简单、高效,有效的减少了使用随机模型所付出的机会成本,帮助管理者快速的调整营销策略。本文的主要贡献如下：1.启发式算法在商业银行活期储蓄领域的应用。将启发式算法首次应用在商业银行活期储蓄领域,对客户进行流失识别、未来交易次数的预测及未来交易金额的预测,其预测结果通过与随机模型BG/NBD和Gamma-Gamma模型实现的预测结果的比较,验证启发式算法在商业银行活期储蓄领域的适用,并再一次说明对客户未来购买行为的预测,启发式算法不逊于随机模型。2.结合商业银行活期储蓄客户的具体数据特征,改进启发式算法对客户未来交易次数E[x]模型的预测。根据商业银行活期储蓄客户的特点,提出符合活期储蓄客户业务背景的预测模型,使改进后的模型更适用于测算我国商业银行活期储蓄客户未来交易次数,进一步提高全部客户总计交易次数的预测准确率。3.跳出常规对客户进行分类的研究思路,直接在客户群中根据计算出的客户终身价值选取20%的优质客户,继而从营销资源合理配置的角度为管理者提供决策依据。更多还原

【Abstract】 Nowadays customers’future purchase behavior has become a hot research in the environment of non-contractual in marketing industry. Academia use stochastic model such as Pareto/NBD, BG/NBD, etc to estimate client’s future purchase behavior. However, due to the complexity of this model, this kind of method is not applied in industry. It is not the scholars’ intention that the theory and practice mismatching. This paper uses heuristic algorithm identify customer churn on the deposit customers in commercial bank for the first time. In comparison with stochastic model, it’s verified that the heuristic algorithm have same effective on the prediction of customers’future purchase behavior, and more simple. This method reduced the opportunity cost effectively and helps managers adjust marketing strategies quickly. It is really a convenient method for managers actual operating. The main contribution of this paperare are as follows:1. The application of heuristic algorithm in the field of current deposit customers in commercial bank. It’s the first time applying the heuristic algorithm in the field of commercial bank’s current deposit. This research consist of the following work:customer churn recognition; forecast the future number of transactions and forecast the amount of future transactions. In comparison with stochastic model, such as BG/NBD and Gamma-Gamma, verified the heuristic algorithm for the prediction of customers’future purchase behavior no less favorable than the complex stochastic models.2. According to the specific data characteristic of saving deposit customer in commercial bank, this paper employ the heuristic algorithm to improve the prediction model of future customer transactions. The research put forward a prediction model that in accordance with the business background of current deposit customers in commercial banks. The improved model is more suitable for measurement of customers’future transactions, and further improves the overall prediction precision of customers. Based on this model we can get a higher forecasting precision about the total number of future purchase.3. According to the calculated customer lifetime value, we selected 20% of the high value customers from the customer base directly, which out of the conventional research ideas about the classification of customers. Then we can provide the basis for managers to make good decision from the perspective of rational allocation of marketing resources.更多还原