Showing papers in "International Journal of Data Analysis Techniques and Strategies in 2008"

Predicting credit card customer churn in banks using data mining

Abstract: In this paper, we solve the customer credit card churn prediction via data mining. We developed an ensemble system incorporating majority voting and involving Multilayer Perceptron (MLP), Logistic Regression (LR), decision trees (J48), Random Forest (RF), Radial Basis Function (RBF) network and Support Vector Machine (SVM) as the constituents. The dataset was taken from the Business Intelligence Cup organised by the University of Chile in 2004. Since it is a highly unbalanced dataset with 93% loyal and 7% churned customers, we employed (1) undersampling, (2) oversampling, (3) a combination of undersampling and oversampling and (4) the Synthetic Minority Oversampling Technique (SMOTE) for balancing it. Furthermore, tenfold cross-validation was employed. The results indicated that SMOTE achieved good overall accuracy. Also, SMOTE and a combination of undersampling and oversampling improved the sensitivity and overall accuracy in majority voting. In addition, the Classification and Regression Tree (CART) was used for the purpose of feature selection. The reduced feature set was fed to the classifiers mentioned above. Thus, this paper outlines the most important predictor variables in solving the credit card churn prediction problem. Moreover, the rules generated by decision tree J48 act as an early warning expert system.

Quality function deployment: a comprehensive literature review

Abstract: In the past few years, various quality standards and quality systems have been attempted for the improvement of the products and services in our lives. One such quality tool which has the ability to generate creative and novel solutions is Quality Function Deployment (QFD). This paper presents a detailed literature review on the topic and application of QFD. This literature review is based on a reference bank of more than 400 QFDs and its allied publications, organisations, software, tools and web sources. The literature review is extended with thorough descriptions of the adopted methodologies, exemplified with an elaborate and categorical application analysis of its varied functional areas, namely, primary, secondary and tertiary fields, industrial, non-industrial and service applications and methodological progressions. The paper concludes with some of the insights gained from a large number of research papers, publications and other available literature.

How problem-solving really works

Abstract: Over the years, many researchers have proposed theoretical models of problem-solving. These models work a problem in a sequential and rational manner. Through our professional experience and an action research study, we discovered fundamental differences between what these models describe and what actually happens when problems are solved in a real-world setting. Assisting with a process improvement experience in a plastics company, we discovered that when a problem is properly identified, problem-solving generally follows the theoretical models. However, when a problem is difficult to identify, problem-solving proceeds in a cyclical and apparently irrational manner. Cyclical problem-solving increases the average time of problem-solving and production cost. The authors find that the relationships among the problem-solving steps are much more complex than implied in existing literature. Incorporating this new understanding into process improvement training reduced the variability of the problem-solving time from 44 to 21 min.

The development of a simulation-based approach to optimise the inventory policy in a single-echelon supply chain: a case study

Abstract: Inventory control is one way to increase profit margins without altering resources. A spreadsheet model was developed for the inventory control of packaging material in a real case study from a fruit juice manufacturing firm. The contribution of this paper is the construction of a spreadsheet model that incorporates a dynamic re-order point policy logic; using the result from the spreadsheet-based approach and analytical approximation method, a best alternate ordering policy for a single-echelon supply chain can be developed by the statistical analysis of the dynamic re-order point, static re-order point and existing ordering policies and the flexibility of the optimal ordering policy was evaluated using a randomly generated demand. The objective of this work is to develop the best ordering policy with a low total inventory cost to ensure a better service efficiency level across a single-echelon supply chain. The reduction in the total inventory costs obtained from the spreadsheet-based approach is compared with the analytical approximation method. This paper provides a basic understanding with respect to the development of an ordering policy for a single-echelon supply chain.

Segmenting the mature travel market by motivation

Abstract: The purpose of this study was to segment mature travellers based on their motivations and to profile the similarities and differences between mature travel market segments according to their sociodemographic and travel-related characteristics. A total of 217 respondents (50 years old and above) in the Upstate area in a southern state in the USA were used in this study. Three types of mature travellers were identified with an exploratory factor analysis and cluster analysis: personal, educational and social travellers. They were significantly different regarding the number of years they had lived in the Upstate.

A composite logistic regression approach for ordinal panel data regression

Abstract: We propose in this article a Composite Logistic Regression (CLR) approach for ordinal panel data regression. The new method transforms the original ordinal regression problem into a number of binary ones. Thereafter, the method of conditional logistic regression (Chamberlain, 1984; Wooldridge, 2001; Hsiao, 2003) can be directly applied. As a result, the new method allows the unobserved subject effects to be correlated with the observed predictors in an arbitrary manner. Computationally, the new method is able to profile out unobserved subject effects in a very neat manner. This not only makes computational implementation very easy but also makes theoretical treatment straightforward. In particular, we show theoretically that the resulting estimator is √n-consistent and asymptotically normal. Both simulations and a real example are reported to demonstrate the usefulness of the new method.

Data-mining application for country segmentation based on the RFM model

Abstract: For effective Customer Relationship Management (CRM), it is important to gather information on customer value. Segmentation is the method of knowing the customers and partitioning a population of customers into smaller groups. This paper develops a novel country segmentation methodology based on Recency (R), Frequency (F) and Monetary value (M) variables. After the variables are calculated, clustering methods (K-means and fuzzy K-means) are used to segment countries and compare the results of these methods by three different criteria. Customers are classified into four tiers: Top-active, Medium-active, New customer and Inactive. Then a customer pyramid is drawn and the customer value is calculated. Consequently, the data are used to analyse the relative profitability of each customer cluster and the proper strategy is determined for them.

Mining fuzzy temporal patterns from process instances with weighted temporal graphs

Abstract: This paper presents an algorithm for mining fuzzy temporal patterns from a given process instance. The fuzzy representation of time intervals embedded between the activities is used for this purpose. Initially, the activities are portrayed with their temporal relationships through temporal graphs and then, the defined data structures are used to retrieve the data suitable for the proposed algorithm. Similar to the familiar k-itemsets and k-dim sequences, their counterparts are introduced in this work. The proposed process-instance level data structure generates an optimum number of temporal itemsets. The proposed algorithm differs from the other existing algorithms on this topic in the representation of the mined data and patterns. An example is provided to demonstrate the algorithm.

The cross-border shoppers' behaviour: the case of Singapore

Abstract: Cross-border shopping has been of interest in the past decade Given the close proximity of Singapore and Johor Bahru (JB) (located at the southern tip of Malaysia), outshopping has become a notable feature of cross-border visits We compared the travel frequencies of 203 Singapore residents who travelled to Malaysia for shopping purposes We also identified and measured the factors that might influence their cross-border shopping behaviour Our results revealed that there was a significant difference in the travel frequencies between the younger and older respondents However, there were no significant differences between males and females and between the low income earners and the high income earners As for the determinants of outshopping behaviour, it was found that marketing mixes and accessibility, as well as economical factors, all had significant and positive influences on cross-border shopping

The effect of team-based performance appraisal on knowledge sharing: constructing and verifying an influencing model

Abstract: Recently, it has been proposed that team-based performance appraisal may promote knowledge sharing in the field of knowledge management. The proposal, however, is theoretically and practically doubtful. As a response, this paper explores how team-based performance appraisal impacts knowledge sharing within teams by developing an influencing model. The factors of team-based performance appraisal that influence knowledge sharing are identified from three aspects: situations, beliefs and motives. The two interactive mechanisms by which team-based performance appraisal can promote knowledge sharing are further established. Moreover, the influencing model is tested by Hierarchical Linear Modelling (HLM) based on data collected from 1128 employees working in 251 teams in 73 organisations in China, and the result demonstrates its validity for organisational management.

Improving public transit access to in-city villages

Abstract: Rapid urbanisation results in the emergence of in-city villages in many developing countries. The lack of public transit access has been a long-standing problem for the residents in such villages. Here we develop a new mathematical model for adding stops onto the existing transit routes passing by the in-city villages, where the stops are decided by the optimisation of an objective function, which consists of the total supplier and user costs. Numerical examples are given to demonstrate the usefulness and efficiency of the proposed model for improving the transit access to the in-city villages.

Generalised percentile ratios as robust measures of labour earnings inequality

Abstract: We propose a class of generalised percentile ratios as an alternative to the P90 / P10 ratio for measuring labour earnings inequality. We show that they are more robust to sampling the variation and rounding error prevalent in interview-based surveys, as demonstrated through a Monte Carlo simulation and with Current Population Survey labour earnings data from 1987 to 2005. We find a smoother upward trend in the P90 / P10 ratio over this period than what is shown with conventionally measured P90 / P10 ratios.