基于MaxDiff和潜在类别分析的移动支付客户细分研究

在竞争激烈的移动支付市场中,对客户进行细分尤为重要。从客户的心理、态度等角度出发,首先采用MaxDiff调查方法测量受访者对移动支付产品的偏好,从而得到MaxDiff数据,再利用潜在类别分析对客户进行分类得到每个细分群体在各个对象上的偏好得分,最后从人口统计学和行为特征方面对细分群体进行验证分析。研究结果显示:客户被分成3个细分群体,分别是"安全和场景"类、"优惠和便捷"类以及"功能和速度"类,它们在移动支付的偏好上有显著差异。这样的细分结果可为移动支付服务商实施精准营销提供决策参考。     

基于改进RFM模型对民航客户的细分研究

近年来航空公司将客户分成不同的群体为了给客户提供差异化服务和有针对性的营销.现有传统的客户细分RFM模型由于存在缺乏科学的指标建立,已无法准确和完整的描述实际情况中客户的细分结果,根据民航客户价值的特点,在传统客户细分的RFM模型上进行改进,创建LRFMC模型,对某航空公司客户采用数据挖掘K-means算法进行聚类分析,得到航空公司旅客分类的结果.依据细分结果对不同旅客提供差异化的服务和有针对性的营销策略,改善民航企业的服务质量,提高民航企业在市场中的竞争力.     

Consensus of partitions : a constructive approach

Given a profile (family) ?? of partitions of a set of objects or items X, we try to establish a consensus partition containing a maximum number of joined or separated pairs in X that are also joined or separated in the profile. To do so, we define a score function, S _?? associated to any partition on X. Consensus partitions for ?? are those maximizing this function. Therefore, these consensus partitions have the median property for the profile and the symmetric difference distance. This optimization problem can be solved, in certain cases, by integer linear programming. We define a polynomial heuristic which can be applied to partitions on a large set of items. In cases where an optimal solution can be computed, we show that the partitions built by this algorithm are very close to the optimum which is reached in practically all the cases, except for some sets of bipartitions.     

Comparing partitions of two sets of units based on the same variables

We propose a procedure based on a latent variable model for the comparison of two partitions of different units described by the same set of variables. The null hypothesis here is that the two partitions come from the same underlying mixture model. We define a method of “projecting” partitions using a supervised classification method: once one partition is taken as a reference; the individuals of the second data set are allocated to the clusters of the reference partition; it gives two partitions of the same units of the second data set: the original and the projected one and we evaluate their difference by usual measures of association. The empirical distributions of the association measures are derived by simulation.     

Semifield flat flocks

Using ‘fusion’ methods on finite semifields, a variety of partitions (flocks) of Segre varieties by caps are obtained. The partitions arise from semifield planes and are thus called “semifield flat flocks”. Furthermore, the finite transitive semifield flat flocks are completely determined.     

A districting procedure for social organizations

The problem dealt with in the paper refers to the decomposition of a geographical region into an unspecified number of non-overlapping regional partitions. Each partition is given by a connected aggregation of indivisible elementary areas and must contain at least one source of a certain social service characterized by assigned capacity and location. Each area is characterized by a service demand and by a measure quantifying its dimension.Two heuristic methods are proposed to obtain a preliminary aggregation of the elementary areas, in order to reduce the computational effort involved in generating the feasible partitions. A fuzzy sets approach is presented for rating and ranking the regional partitions. This approach seems to be particularly suited to planning problems, when some kind of compromise between qualitative and quantitative issues must be sought. A health-care districting problem is then discussed as a case study.     

Customer lifetime value: stochastic optimization approach

Since the early 1980s, the concept of relationship marketing has been becoming important in general marketing, especially in the area of direct and interactive marketing. The core of relationship marketing is the maintenance of long-term relationships with the customers. However, the relationship marketing is costly and therefore, the determination of the customer lifetime value (CLV) is an important element in making strategic decisions in both advertising and promotion. In this paper, we propose a stochastic dynamic programming model with a Markov chain for the optimization of CLV. Both cases of infinite horizon and finite horizon are discussed. The model is then applied to practical data of a computer service company.     

N-graphs,Modular Sidon and Sum-Free Sets,and Partition Identities

Using a new graphical representation for partitions, the author obtains a family of partition identities associated with partitions into distinct parts of an arithmetic progression, or, more generally, with partitions into distinct parts of a set that is a finite union of arithmetic progressions associated with a modular sum-free Sidon set. Partition identities are also constructed for sets associated with modular sum-free sets.     

Customer choice prediction based on transfer learning

Choice behaviour prediction is valuable for developing suitable customer segmentation and finding target customers in marketing management. Constructing good choice models for choice behaviour prediction usually requires a sufficient amount of customer data. However, there is only a small amount of data in many marketing applications due to resource constraints. In this paper, we focus on choice behaviour prediction with a small sample size by introducing the idea of transfer learning and present a method that is applicable to choice prediction. The new model called transfer bagging extracts information from similar customers from different areas to improve the performance of the choice model for customers of interest. We illustrate an application of the new model for customer mode choice analysis in the long-distance communication market and compare it with other benchmark methods without information transfer. The results show that the new model can provide significant improvements in choice prediction.     

基于模糊聚类算法的客户细分研究

以客户关系为导向的服务经济时代,客户细分是企业判断具有相似特征的客户,并以此为基础有针对性的提供产品及服务的最为有效的手段.利用数据挖掘技术,将模糊聚类算法应用于客户细分中,以此寻找具有相似特征及价值的客户,并在电信行业中进行实证应用.     

Nash game model for optimizing market strategies,configuration of platform products in a Vendor Managed Inventory (VMI) supply chain for a product family

This paper discusses how a manufacturer and its retailers interact with each other to optimize their product marketing strategies, platform product configuration and inventory policies in a VMI (Vendor Managed Inventory) supply chain. The manufacturer procures raw materials from multiple suppliers to produce a family of products sold to multiple retailers. Multiple types of products are substitutable each other to end customers. The manufacturer makes its decision on raw materials’ procurement, platform product configuration, product replenishment policies to retailers with VMI, price discount rate, and advertising investment to maximize its profit. Retailers in turn consider the optimal local advertising investments and retail prices to maximize their profits. This problem is modeled as a dual simultaneous non-cooperative game (as a dual Nash game) model with two sub-games. One is between the retailers serving in competing retail markets and the other is between the manufacturer and the retailers. This paper combines analytical, iterative and GA (genetic algorithm) methods to develop a game solution algorithm to find the Nash equilibrium. A numerical example is conducted to test the proposed model and algorithm, and gain managerial implications.     

Building initial partitions through sampling techniques

A variety of iterative clustering algorithms require an initial partition of a dataset as an input parameter. As a rule a good choice of the initial partition is essential for building a high quality final partition. In this note, we generate initial partitions by using small samples of the data. Numerical experiments with k-means like clustering algorithms are reported.     

Compositions of n as alternating sequences of weakly increasing and strictly decreasing partitions

Compositions and partitions of positive integers are often studied in separate frameworks where partitions are given by q-series generating functions and compositions exhibiting specific patterns are designated by generating functions for these patterns. Here, we view compositions as alternating sequences of weakly increasing and strictly decreasing partitions (i.e. alternating blocks). We obtain generating functions for the number of such partitions in terms of the size of the composition, the number of parts and the total number of “valleys” and “peaks”. From this, we find the total number of “peaks” and “valleys” in the composition of n which have the mentioned pattern. We also obtain the generating function for compositions which split into just two partition blocks. Finally, we obtain the two generating functions for compositions of n that start either with a weakly increasing partition or a strictly decreasing partition.     

基于排队系统的偏好差异性顾客服务策略分析

在服务商为不同偏好顾客提供差异性服务的排队系统中,其收益受到需求规模、顾客类型等因素影响,如何制定策略才能达到服务特定类型顾客并获取最大收益的目的？本文以服务时间与效果正相关的顾客密集型服务为背景,基于M/M/1模型对服务商为两种偏好顾客提供两种服务的情景进行研究,并将从偏好服务所获基准效用相对较低、较高的顾客分别定义为1类、2类顾客。结果表明,①随着潜在顾客规模的增加,最优策略所适用的3种基本策略依次为公平策略、优先服务2类顾客的策略以及仅服务2类顾客的排他性策略;②3种基本策略的适用范围仅与1类顾客的服务选择标准相关;③在优先服务某类顾客的策略中,有效到达速率及服务速度由另一类顾客的特征决定。     

Orthogonal Partitions in Designed Experiments

A survey is given of the statistical theory of orthogonal partitions on a finite set. Orthogonality, closure under suprema, and one trivial partition give an orthogonal decomposition of the corresponding vector space into subspaces indexed by the partitions. These conditions plus uniformity, closure under infima and the other trivial partition give association schemes. Examples covered by the theory include Latin squares, orthogonal arrays, semilattices of subgroups, and partitions defined by the ancestral subsets of a partially ordered set (the poset block structures).Isomorphism, equivalence and duality are discussed, and the automorphism groups given in some cases. Finally, the ideas are illustrated by some examples of real experiments.Dedicated to Hanfried Lenz on the occasion of his 80th birthday.     

Price and speed decisions in customer-intensive services with two classes of customers

In many customer-intensive services, the perceived quality of service decreases in the speed of service. Usually, an increase in service speed induces different marginal reductions in quality for heterogeneous customers. To bring insight into the managerial implications of this difference, we classify customers in terms of intensity parameters, and investigate the behavior of each class of customers in a queueing framework. The optimal service speed and price are derived to maximize service provider’s revenue. Our results demonstrate that no class is always attractive to the provider, and thus there are usually several combinations of service price and speed reaching the same maximal revenue. Moreover, under some mild conditions, the provider could gain more revenue by treating different classes with discrimination than by adopting uniform treatment.     

Orthogonal partitions in designed experiments

A survey is given of the statistical theory of orthogonal partitions on a finite set. Orthogonality, closure under suprema, and one trivial partition give an orthogonal decomposition of the corresponding vector space into subspaces indexed by the partitions. These conditions plus uniformity, closure under infima and the other trivial partition give association schemes. Examples covered by the theory include Latin squares, orthogonal arrays, semilattices of subgroups, and partitions defined by the ancestral subsets of a partially ordered set (the poset block structures).Isomorphism, equivalence and duality are discussed, and the automorphism groups given in some cases. Finally, the ideas are illustrated by some examples of real experiments.Dedicated to Hanfried Lenz on the occasion of his 80th birthday.     

On Potts Model Clustering,Kernel K-Means and Density Estimation

Many clustering methods, such as K -means, kernel K -means, and MNcut clustering, follow the same recipe: (i) choose a measure of similarity between observations; (ii) define a figure of merit assigning a large value to partitions of the data that put similar observations in the same cluster; and (iii) optimize this figure of merit over partitions. Potts model clustering represents an interesting variation on this recipe. Blatt, Wiseman, and Domany defined a new figure of merit for partitions that is formally similar to the Hamiltonian of the Potts model for ferromagnetism, extensively studied in statistical physics. For each temperature T, the Hamiltonian defines a distribution assigning a probability to each possible configuration of the physical system or, in the language of clustering, to each partition. Instead of searching for a single partition optimizing the Hamiltonian, they sampled a large number of partitions from this distribution for a range of temperatures. They proposed a heuristic for choosing an appropriate temperature and from the sample of partitions associated with this chosen temperature, they then derived what we call a consensus clustering: two observations are put in the same consensus cluster if they belong to the same cluster in the majority of the random partitions. In a sense, the consensus clustering is an “average” of plausible configurations, and we would expect it to be more stable (over different samples)than the configuration optimizing the Hamiltonian.

The goal of this article is to contribute to the understanding of Potts model clustering and to propose extensions and improvements: (1) We show that the Hamiltonian used in Potts model clustering is closely related to the kernel K -means and MNCutcriteria. (2) We propose a modification of the Hamiltonian penalizing unequal clustersizes and show that it can be interpreted as a weighted version of the kernel K -meanscriterion. (3) We introduce a new version of the Wolff algorithm to simulate configurations from the distribution defined by the penalized Hamiltonian, leading to penalized Potts model clustering. (4) We note a link between kernel based clustering methods and nonparametric density estimation and exploit it to automatically determine locally adaptive kernel bandwidths. (5) We propose a new simple rule for selecting a good temperature T.

As an illustration we apply Potts model clustering to gene expression data and compare our results to those obtained by model based clustering and a nonparametric dendrogram sharpening method.     

Optimal threshold analysis of segmentation methods for identifying target customers

In CRM (Customer Relationship Management), the importance of a segmentation method for identifying good customers has been increasing. For evaluation of different segmentation methods, Accuracy often plays a key role. This indicator, however, cannot distinguish two types of errors. The purpose of this paper is to overcome this pitfall by introducing two different indicators: Recall and Precision. Assuming that a promotion is addressed exclusively to the selected target customers, the financial effectiveness of the underlying segmentation method is expressed as a function of Recall and Precision. An optimization problem is then formulated so as to maximize the financial measure by finding the optimal threshold level in terms of the severeness for estimating the target set. By introducing a functional form which represents correctness and mistakes about the target set, the unique optimal solution is derived explicitly. The proposed approach is validated by using real customer purchase data.     

A duopoly model with heterogeneous congestion-sensitive customers

This paper analyzes a model with two firms (providers), and two classes of customers. These customers classes are characterized by their attitude towards ‘congestion’ (caused by other customers using the same resources); a firm is selected on the basis of both the prices charged by the firms, and the ‘congestion levels’. The model can be represented by a two-stage game: in the first providers set their prices, whereas in the second the customers choose the provider (or to not use any service at all) for given prices. We explicitly allow the providers to split their resources, in order to serve more than just one market segment. This enables us to further analyze the Paris metro pricing (Pmp) proposal for service differentiation in the Internet.