Showing papers by "Jian Zhou published in 2020"

Market segmentation using high-dimensional sparse consumers data

Abstract: Good segmentation contributes towards a better understanding of the market and customer demands. This study aims to develop a new methodological approach, integrating “Recency, Frequency and Monetary” with the sparse K-means clustering algorithm of Witten and Tibshirani (2010). The proposed approach is suitable for handling large, high-dimensional and sparse consumer data. Drawing on the proposed methodology, alongside data collection from the Chinese mobile telecommunications market, and considering specific services, our treatment is further assessed empirically and appears to provide robust results when compared to the Dolnicar, Kaiser, Lazarevski and Leisch (2012) biclustering of customers method. Following the attainment of a clear and robust market segmentation structure, our theoretical treatment and its empirical analysis provide a useful tool and valid methodology for marketers, and decision makers in general, to accurately determine the most profitable market segments.

A Novel Closed-Loop Supply Chain Network Design Considering Enterprise Profit and Service Level

Abstract: Increasing concerns for sustainable development have motivated the study of closed-loop supply chain network design from a multidimensional perspective. To cope with such issues, this paper presents a general closed-loop supply chain network comprising various recovery options and further formulates a multi-objective mixed-integer linear programming model considering enterprise profit and service level simultaneously. Within this model, market segmentation is also considered to meet real-world operating conditions. Moreover, an e -constraint method and two interactive fuzzy approaches are applied to find a global optimum for this model together with the decisions on the numbers, locations, and capacities of the facilities, as well as the material flow through the network. Ultimately, numerical experiments are conducted to demonstrate the viability and effectiveness of both the proposed model and solution approaches.

Fuzzy belief propagation in constrained Bayesian networks with application to maintenance decisions

Abstract: Bayesian networks have been widely applied to domains such as medical diagnosis, fault analysis, and preventative maintenance. In some applications, because of insufficient data and the complexity ...

A biclustering-based heterogeneous customer requirement determination method from customer participation in product development

Abstract: Timely identification of heterogeneous customer requirements serves as a vital step for a company to formulate product strategies to meet the diverse and changing needs of its customers. By relaxing the search for global patterns in classical clustering, we propose a biclustering-based method, BiHCR, to identify heterogeneous customer requirements from the perspective of local patterns detection. Specifically, conforming to customers’ attitudes toward products derived from customer participation, we first transform the original data matrix with customers as rows and customer requirements as columns into a binary matrix. Then, by combining the two significant biclustering algorithms, Bimax and RepBimax, we design BiHCR to identify the biclusters embedded in the binary matrix to improve the detection results from the larger biclusters and their overlaps. Furthermore, the empirical case of smartphone development in a Chinese company verifies that BiHCR can identify homogeneous subgroups of customers with similar requirements without redundant noise compared with Bimax. Additionally, in contrast to RepBimax, our proposed BiHCR can also detect the intractable overlapping biclusters in the binary matrix used to describe the heterogeneity of customer requirements. Since the process of customer participation in product development gradually became a dominant approach to collecting customer requirements information for many industries, a conceptual framework of customer requirements identification is constructed and the detailed steps are clarified for manufacturers.

QFD-based product design for multisegment markets: a fuzzy chance-constrained programming approach

Abstract: Modern companies are moving rapidly toward more customer-oriented approaches for new product design and development. In real-world applications, customer requirements are more diverse and rather heterogeneous in practice, which requires the targeted market to be multisegmented. In this article, employing the quality function deployment (QFD) and assuming that parameter uncertainty exists, we propose a new fuzzy QFD model for use in multisegment markets, to search for optimal engineering characteristics by maximizing the overall customer satisfaction score. Under that framework, and when the credibility measure is employed, we adopt a fuzzy chance-constrained programming setting to convert the fuzzy model into its crisp counterpart. Afterwards, an analytical method based on fuzzy operational law is proposed to further transform the chance-constrained programming into an equivalent explicit model that can be solved directly by a standard commercial software. For a targeted group of customers, a new yoghurt design is presented to illustrate the effectiveness and superiority of our treatment. We report that our methodology not only permits the selection of the preferred solution based on an acceptable level of risk, which is very important in practice, but also guarantees higher customer satisfaction scores. This fact extends the fuzzy QFD methodology to a wider range of applications.

Fuzzy Bi-Objective Closed-Loop Supply Chain Network Design Problem with Multiple Recovery Options

Abstract: A network design of a closed-loop supply chain (CLSC) with multiple recovery modes under fuzzy environments is studied in this article, in which all the cost coefficients (e.g., for facility establishment, transportation, manufacturing and recovery), customer demands, delivery time, recovery rates and some other factors that cannot be precisely estimated while designing are modeled as triangular fuzzy numbers. To handle these uncertain factors and achieve a compromise between the two conflicting objectives of maximizing company profit and improving customer satisfaction, a fuzzy bi-objective programming model and a corresponding two-stage fuzzy interactive solution method are presented. Applying the fuzzy expected value operator and fuzzy ranking method, the fuzzy model is transformed into a deterministic counterpart. Subsequently, Pareto optimal solutions are determined by employing the fuzzy interactive solution method to deal with the conflicting objectives. Numerical experiments address the efficiency of the proposed model and its solution approach. Furthermore, by comparing these results with the CLSC network design in deterministic environments, the benefits of modeling the CLSC network design problem with fuzzy information are highlighted.