面向分类属性数据的一种改进相异性度量及其在客户细分中的应用*

具有层次结构的分类属性在客户细分应用中广泛存在。针对传统相异性度量无法准确反映决策者在与细分目标相关的决策指标上的偏好信息,提出一种改进的距离层次并给出使用该度量,基于聚类分析的客户细分基本流程。该度量利用距离层次计算各分类属性值概念间的相异性,同时引入指标距离的概念描述对于特定指标,决策者在不同分类属性值上的偏好,结合模糊相似优先比决策方法和树的广度优先遍历计算不同分类属性值间的指标距离,最后通过将所求得的概念距离和指标距离进行加权求和以更全面地度量不同分类属性值间的相异性。对陕西省电力公司工业客户进行细分实验的结果表明：与传统距离层次相比,采用改进相异性度量能提高聚类质量和细分结果的可解释性。     

Customer segmentation issues and strategies for an automobile dealership with two clustering techniques

Companies can use customer segmentation to group customers with similar characteristics together and identify the differences between groups to develop marketing strategies. This study investigates the problem of customer segmentation in relation to automotive customer relationship management and presents a real case study of an automobile dealer in Taiwan. Although several past studies have adopted different clustering techniques with which to group customer attributes, few have simultaneously considered customer transaction behaviour and customer satisfaction variables. In addition, most previous work has used only a single clustering method for customer segmentation, which results in unreliable results and leads to inadequate marketing decisions. Therefore, in this study, we consider two clustering techniques, k‐means and expectation maximization, and compare their results for correctness. The experimental results show that four customer groups are identified with both clustering methods: loyal, potential, VIP and churn customer groups. Based on the segmentation results, several customized marketing strategies aimed at each of the four customer groups are suggested to improve the quality of services for effective customer relationship management.     

客户导向目录分割问题的改进算法

客户导向目录分割问题假设顾客至少对目录中一定数量的商品感兴趣,计算目录覆盖的顾客数量,据此评估目录分割结果. 现有的分割算法为了保证目录尽可能多的覆盖顾客,而忽略了目录分割结果的效用. 针对该问题,本文构建一种新的数据存储结构CFP-Tree用于存储顾客交易数据,并提出一种新的算法Effective-Cover解决目录分割问题. 该算法使用树深度遍历法选择目录产品. 实验结果表明,该算法能够获得更好的目录分割结果.     

A transaction pattern analysis system based on neural network

Customer segmentation is a key element for target marketing or market segmentation. Although there are quite a lot of ways available for segmentation today, most of them emphasize numeric calculation instead of commercial goals. In this study, we propose an improved segmentation method called transaction pattern based customer segmentation with neural network (TPCSNN) based on customer’s historical transaction patterns. First of all, it filters transaction data from database for records with typical patterns. Next, it reduces inter-group correlation coefficient and increases inner cluster density to achieve customer segmentation by iterative calculation. Then, it utilizes neural network to dig patterns of consumptive behaviors. The results can be used to segment new customers. By this way, customer segmentation can be implemented in very short time and costs little. Furthermore, the results of segmentation are also analyzed and explained in this study.     

Integration of self-organizing feature maps neural network and genetic K-means algorithm for market segmentation

This study is dedicated to proposing a novel two-stage method, which first uses Self-Organizing Feature Maps (SOM) neural network to determine the number of clusters and the starting point, and then uses genetic K-means algorithm to find the final solution. The results of simulated data via a Monte Carlo study show that the proposed method outperforms two other methods, K-means and SOM followed by K-means (Kuo, Ho & Hu, 2002a), based on both within-cluster variations (SSW) and the number of misclassification. In order to further demonstrate the proposed approach's capability, a real-world problem of the fright transport industry market segmentation is employed. A questionnaire is designed and surveyed, after which factor analysis extracts the factors from the questionnaire items as the basis of market segmentation. Then the proposed method is used to cluster the customers. The results also indicate that the proposed method is better than the other two methods     

A cross-national market segmentation of online game industry using SOM

To compete successfully in today's global online game markets, a cross-national analysis for market segmentation is becoming a more important issue, by which companies are able to understand their domestic and foreign loyal customers and concentrate their limited resources into the target customers. However, previous research methodologies for market segmentation were difficult to be conducted on a cross-national analysis because they were performed within a nation. Additionally, the traditional clustering methodologies have not provided a unique clustering nor determined the precise number of clusters.

The purpose of our research is to develop a new methodology for cross-national market segmentation. We propose a two-phase approach (TPA) integrating statistical and data mining methods. The first phase is conducted by a statistical method (MCFA: multi-group confirmatory factor analysis) to test the difference between national clustering factors. The second phase is conducted by a data mining method (a two-level SOM) to develop the actual clusters within each nation. A two-level SOM is useful to effectively reduce the complexity of the reconstruction task and noise. Especially, our research tested the model with Korean and Japanese online game users because they are the frontier of global online game industries.     

Data mining-based hierarchical transaction model for multi-level consistency management in large-scale replicated databases

Scalability and availability in a large-scale distributed database is determined by the consistency strategies used by the transactions. Most of the big data applications demand consistency and availability at the same time. However, a suitable transaction model that handles the trade-obetween availability and consistency is presently lacking. In this article, we have proposed a hierarchical transaction model that supports multiple consistency levels for data items in a large-scale replicated database. The data items have been classified into different categories based on their consistency requirement, computed using a data mining algorithm. Thereafter, these have been mapped to the appropriate consistency level in the hierarchy. This allows parallel execution of several transactions belonging to each level. The topmost level called the Serializable (SR) level follows strong consistency applicable to data items that are mostly read and updated both. The next level of consistency, Snapshot Isolation (SI), maps to data items which are mostly read and demand unblocking read. Data items which are mostly updated do not follow strict consistent snapshot and have been mapped to the next lower level called Non- monotonic Snapshot Isolation (NMSI). The lowest level in the hierarchy correspond to data items for which ordering of operations does not matter. This level is called the Asynchronous (ASYNC) level. We have tested the proposed transaction model with two different workloads on a test-bed designed following the TPC-C benchmark schema. The performance of the proposed model has been evaluated against other transaction models that support single consistency policy. The proposed model has shown promising results in terms of transaction throughput, commit rate and average latency.     

Intelligent value-based customer segmentation method for campaign management: A case study of automobile retailer

Most marketers have difficulty in identifying the right customers to engage in successful campaigns. So far, customer segmentation is a popular method that is used for selecting appropriate customers for a launch campaign. Unfortunately, the link between customer segmentation and marketing campaign is missing. Another problem is that database marketers generally use different models to conduct customer segmentation and customer targeting. This study presents a novel approach that combines customer targeting and customer segmentation for campaign strategies. This investigation identifies customer behavior using a recency, frequency and monetary (RFM) model and then uses a customer life time value (LTV) model to evaluate proposed segmented customers. Additionally, this work proposes using generic algorithm (GA) to select more appropriate customers for each campaign strategy. To demonstrate the efficiency of the proposed method, this work performs an empirical study of a Nissan automobile retailer to segment over 4000 customers. The experimental results demonstrate that the proposed method can more effectively target valuable customers than random selection.     

Integrated rough VIKOR for customer-involved design concept evaluation combining with customers’ preferences and designers’ perceptions

Customer-involved design concept evaluation (DCE) allows customers to take part in evaluating the design alternatives to get more popular design concept. Traditional customer-involved DCE methods still focus on the collection of customer responses and only consider cost and benefit characteristics of design criteria in multi-criteria decision-making (MCDM) based evaluation process. Few studies have customized the decision-making algorithms specifically aimed at customers’ preferences. This paper further explores the customers’ influences in the early stages of the product design development, and proposes a new rough number based MCDM model (i.e., VIKOR) incorporating customers’ preferences for design specifications along with designers’ perceptions for the characteristics of design criteria (cost and benefit) to perform concept evaluation under subjective environment, and this proposed method is named as integrated rough VIKOR (IR-VIKOR). The objective of this study is to identify the best design concept which maximizes the satisfactions of expectations from most customers as well as conforms to the characteristics of design criteria. Firstly, Shannon entropy is used to obtain the weightings and relative importance ratings of design criteria from the customers’ preferences. Secondly, the customers’ preferences for design attribute values, the importance ratings of design criteria and the characteristics of design criteria are combined together to define the ideal solutions to calculate the rough evaluation index of each design alternative in IR-VIKOR, and finally the ranking result is provided by IR-VIKOR to determine the best design concept. A practical design example is introduced to illustrate the evaluation process of this proposed method, and the empirical comparisons are further carried out to validate its superiority for DCE. Through the sensitivity analysis experiments including i) inside IR-VIKOR, and ii) between IR-VIKOR and other classical MCDM methods, the proposed method is proved to be a reliable and feasible customer-involved DCE approach.     

Which is more important in Internet shopping,perceived price or trust?

Price and trust are considered to be two important factors that influence customer purchasing decisions in Internet shopping. This paper examines the relative influence they have on online purchasing decisions for both potential and repeat customers. The knowledge of their relative impacts and changes in their relative roles over customer transaction experience is useful in developing customized sales strategies to target different groups of customers. The results of this study revealed that perceived trust exerted a stronger effect than perceived price on purchase intentions for both potential and repeat customers of an online store. The results also revealed that perceived price exerted a stronger influence on purchase decisions of repeat customers as compared to that of potential customers. Perceived trust exerted a stronger influence on purchase decisions of potential customers as compared to that of repeat customers.     

Customer data mining for lifestyle segmentation

A good relationship between companies and customers is a crucial factor of competitiveness. Market segmentation is a key issue for companies to develop and maintain loyal relationships with customers as well as to promote the increase of company sales. This paper proposes a method for market segmentation in retailing based on customers’ lifestyle, supported by information extracted from a large transactional database. A set of typical shopping baskets are mined from the database, using a variable clustering algorithm, and these are used to infer customers lifestyle. Customers are assigned to a lifestyle segment based on their purchases history. This study is done in collaboration with an European retailing company.     

一种高效的多层和概化关联规则挖掘方法

通过对分类数据的深入研究,提出了一种高效的多层关联规则挖掘方法:首先,根据分类数据所在的领域知识构建基于领域知识的项相关性模型DICM(domain knowledge-based item correlation model),并通过该模型对分类数据的项进行层次聚类;然后,基于项的聚类结果对事务数据库进行约简划分;最后,将约简划分后的事务数据库映射至一种压缩的AFOPT树形结构,并通过遍历AFOPT树替代原事务数据库来挖掘频繁项集.由于缩小了事务数据库规模,并采用了压缩的AFOPT结构,所提出的方法有效地节省了算法的I/O时间,极大地提升了多层关联规则的挖掘效率.基于该方法,给出了一种自顶向下的多层关联规则挖掘算法TD-CBP-MLARM和一种自底向上的多层关联规则挖掘算法BU-CBP-MLARM.此外,还将该挖掘方法成功扩展至概化关联规则挖掘领域,提出了一种高效的概化关联规则挖掘算法CBP-GARM.通过大量人工随机生成数据的实验证明,所提出的多层和概化关联规则挖掘算法不仅可以确保频繁项集挖掘结果的正确性和完整性,还比现有同类最新算法具有更好的挖掘效率和扩展性.     

A two-stage clustering method to analyze customer characteristics to build discriminative customer management: A case of textile manufacturing business

In order to obtain comprehensive information about customers, this study aims to use a systematized analytic method to examine customers. This study uses LRFM customer relationship model, which consists of four dimensions: relation length (L), recent transaction time (R), buying frequency (F), and monetary (M), to carry out customer clusters. We proceed with this clustering analysis to classify customers in order to set discriminative marketing strategies. In addition, this study further employed a cross analysis over three predetermined dimensions: area, sales, and new/old characteristics to enhance the clustering analysis. The results obtained from the real textile business show that the customer groups formed using the four-factor (LRFM) clustering all has statistical significant differences, and with meaningful explanations in terms of marketing strategy. Thus, this study considers useful for discriminative customer relationship management.     

Clustering and selecting suppliers based on simulated annealing algorithms

This study proposes two optimization mathematical models for the clustering and selection of suppliers. Model 1 performs an analysis of supplier clusters, according to customer demand attributes, including production cost, product quality and production time. Model 2 uses the supplier cluster obtained in Model 1 to determine the appropriate supplier combinations. The study additionally proposes a two-phase method to solve the two mathematical models. Phase 1 integrates k-means and a simulated annealing algorithm with the Taguchi method (TKSA) to solve for Model 1. Phase 2 uses an analytic hierarchy process (AHP) for Model 2 to weight every factor and then uses a simulated annealing algorithm with the Taguchi method (ATSA) to solve for Model 2. Finally, a case study is performed, using parts supplier segmentation and an evaluation process, which compares different heuristic methods. The results show that TKSA+ATSA provides a quality solution for this problem.     

Satisfaction Benchmarking and Customer Classification: An Application to the Branches of a Banking Organization

Banking services constitute a highly competitive market and indicate a representative example of customer–oriented organizations. For this reason, customer satisfaction is of vital importance, offering a quantitative measure for current and future performance of these organizations. On the other hand, classifying customers according to their satisfaction behavior may indicate different client clusters with distinctive preferences and expectations. This approach serves the development of a truly customer–focussed culture by determining a set of improvement strategies that best fit the different customer segments. In this paper, the multi–criteria method MUSA is implemented in order to measure and analyze customer satisfaction in different branches of a banking organization. These results are also used to benchmark these branches according to the provided services. Moreover, segmentation analysis is performed in order to identify the different groups of customers and estimate the homogeneity of preferences in distinct customer segments. Data are based on a pilot customer–satisfaction survey, while the most important results are focussed on the determination of the critical service dimensions.     

Critical purchasing incidents in e-business

This research utilized the critical incident technique (CIT) to identify factors influencing customer satisfaction and retention of customers participating in e-commerce transactions. Customers were asked in telephone interviews to discuss both particularly satisfying and dissatisfying (or critical) incidents they had experienced when using web sites to conduct transactions. Each customer also provided demographic information, rated their satisfaction with the experience and ecommerce provider, and was asked how often they purchased products from the provider prior to and after the incident. Analysis revealed 662 citations by customers of items contributing to either positive or negative experiences. Exploratory text-mining analysis revealed that the majority of positive items pertained to the transaction (38%), product (25%), or website (23%), and to customer support (35%) or the transaction for negative items (30%). Approximately 86% of customers citing positive items said they were very likely to use the e-commerce site again, compared to 22% of customers citing negative items. Customer support and user experience both seemed to play an important mediating role on the criticality of negative incidents. Over 70% of first time users and customers who said customer support ignored or refused their requests for assistance said they were unlikely to return to the site, compared to 20% when customer support was said to be responsive. Correlation analysis confirmed that negative incidents tended to be more critical than positive ones, and more so for first time customers.     

基于字簇的多模型中文分词方法研究

字标注分词方法是当前中文分词领域中一种较为有效的分词方法,但由于中文汉字本身带有语义信息,不同字在不同语境中其含义与作用不同,导致每个字的构词规律存在差异。针对这一问题,提出了一种基于字簇的多模型中文分词方法,首先对每个字进行建模,然后对学习出的模型参数进行聚类分析形成字簇,最后基于字簇重新训练模型参数。实验结果表明,该方法能够有效地发现具有相同或相近构词规律的字簇,很好地区别了同类特征对不同字的作用程度。     

Mining interval sequential patterns

The main task of mining sequential patterns is to analyze the transaction database of a company in order to find out the priorities of items that most customers take when consuming. In this article, we propose a new method—the ISP Algorithm. With this method, we can find out not only the order of consumer items of each customer, but also offer the periodic interval of consumer items of each customer. Compared with other previous periodic association rules, the difference is that the period the algorithm provides is not the repeated purchases in a regular time, but the possible repurchases within a certain time frame. The algorithm utilizes the transaction time interval of individual customers and that of all the customers to find out when and who will buy goods, and what items of goods they will buy. © 2005 Wiley Periodicals, Inc. Int J Int Syst 20: 359–373, 2005.     

Benefits of quantile regression for the analysis of customer lifetime value in a contractual setting: An application in financial services

The move towards a customer-centred approach to marketing, coupled with the increasing availability of customer transaction data, has led to an interest in understanding and estimating customer lifetime value (CLV). Several authors point out that, when evaluating customer profitability, profitable customers are rare compared to the unprofitable ones. In spite of this, most authors fail to recognize the implications of these skewed distributions on the performance of models they use. In this study, we propose analyzing CLV by means of quantile regression. In a financial services application, we show that this technique provides management more in-depth insights into the effects of the covariates that are missed with linear regression. Moreover, we show that in the common situation where interest is in a top-customer segment, quantile regression outperforms linear regression. The method also has the ability of constructing prediction intervals. Combining the CLV point estimate with the prediction intervals leads to a new segmentation scheme that is the first to account for uncertainty in the predictions. This segmentation is ideally suited for managing the portfolio of customers.     

Group RFM analysis as a novel framework to discover better customer consumption behavior

The RFM model provides an effective measure for customers’ consumption behavior analysis, where three variables, namely, consumption interval, frequency, and money amount are used to quantify a customer’s loyalty and contribution. Based on the RFM value, customers can be clustered into different groups and the group information is very useful in market decision making. However, most previous works completely left out important characteristics of purchased products, such as their prices and lifetimes, and apply the RFM measure on all of a customer’s purchased products. This renders the calculation of the RFM value unreasonable or insignificant for customer analysis. In this paper, we propose a new framework called GRFM (for group RFM) analysis to alleviate the problem. The new measure method takes into account the characteristics of the purchased items so that the calculated the RFM value for the customers are strongly related to their purchased items and can correctly reflect their actual consumption behavior. Moreover, GRFM employs a constrained clustering method PICC (for Purchased Items-Constrained Clustering) that could base on a cleverly designed purchase pattern table to adjust original purchase records to satisfy various clustering constraints as well as to decrease re-clustering time. The GRFM allows a customer to belong to different clusters, and thus to be associated with different loyalties and contributions with respect to different characteristics of purchased items. Finally, the clustering result of PICC contains extra information about the distribution status inside each cluster that could help the manager to decide when is most proper to launch a specific sales promotion campaign. Our experiments have confirmed the above observations and suggest that GRFM can play an important role in building a personalized purchasing management system and an inventory management system.