Modeling of individual customer delivery satisfaction: an AutoML and multi-agent system approach

Date13 May 2019
Published date13 May 2019
DOIhttps://doi.org/10.1108/IMDS-07-2018-0279
Pages840-866
AuthorW.M. Wang,J.W. Wang,A.V. Barenji,Zhi Li,Eric Tsui
Subject MatterInformation & knowledge management,Information systems,Data management systems,Knowledge management,Knowledge sharing,Management science & operations,Supply chain management,Supply chain information systems,Logistics,Quality management/systems
Modeling of individual customer
delivery satisfaction: an AutoML
and multi-agent system approach
W.M. Wang, J.W. Wang, A.V. Barenji and Zhi Li
School of Electromechanical Engineering, Guangdong University of Technology,
Guangzhou, China, and
Eric Tsui
Department of Industrial and Systems Engineering,
Knowledge Management and Innovation Research Centre,
The Hong Kong Polytechnic University, Kowloon, Hong Kong
Abstract
Purpose The purposeof this paper is to propose an automatedmachine learning (AutoML)and multi-agent
system approach to improve overall product delivery satisfaction under limited resources.
Design/methodology/approach An AutoML method is purposed to model delivery satisfaction of
individual customer, and a heuristic method and multi-agent system are proposed to improve overall
satisfaction under limited processing capability. A series of simulation experiments have been conducted to
illustrate the effectiveness of the proposed methodology.
Findings The simulated results show that the proposed method can effectively improve overall delivery
satisfaction, especially when the demand of customer orders is highly fluctuating and when the customer
satisfaction models are highly diversified.
Practical implications The proposed framework provides a more dynamic and continuously improving
way to model delivery satisfaction of individual customer, thereby supports companies to provide
personalized services and develop scalable and flexible business at a lower cost, and ultimately improves the
overall quality, efficiency and effectiveness of delivery services.
Originality/value The proposed methodology utilizes AutoML and multi-agent system to model
individual customer delivery satisfaction and improve the overall satisfaction. It can cooperate with the
existing delivery resource planning methods to further improve customer delivery satisfaction. The authors
propose an AutoML approach to model individual customer delivery satisfaction, which enables continuous
update and improvements. The authors propose multi-agent system and a heuristic method to improve
overall delivery satisfaction. The numerical results show that the proposed method can improve overall
delivery satisfaction with limited processing capability.
Keywords Multi-agent system, AutoML, Customer delivery satisfaction, Delivery optimization,
Product delivery
Paper type Research paper
1. Introduction
The internet has caused fundamental changes in the retail business. Research shows that
customer satisfaction has a positive effect on consumer spending in e-commerce (Nisar and
Prabhakar, 2017). There are many factors affecting customer satisfaction in the e-commerce
market, such as branding, product quality and service quality. Among them, quality of
service is a key factor affecting online spending (Nisar and Prabhakar, 2017). It includes
elements such as delivery and shipping time, costs and choice, risk reduction, payment
system and security, communication and many other elements (Chintagunta et al., 2012).
Industrial Management & Data
Systems
Vol. 119 No. 4, 2019
pp. 840-866
© Emerald PublishingLimited
0263-5577
DOI 10.1108/IMDS-07-2018-0279
Received 3 July 2018
Revised 10 September 2018
22 September 2018
Accepted 3 October 2018
The current issue and full text archive of this journal is available on Emerald Insight at:
www.emeraldinsight.com/0263-5577.htm
This work was supported by the National Natural Science Foundation of China under Grant
No. 51405089; the Science and Technology Planning Project of Guangdong Province under Grant
Nos 2015B010131008 and 2015B090921007; China Postdoctoral Science Foundation under
Grant Nos 2018M630928 and 2018M633008; and Natural Science Foundation of Guangdong
Province under Grant No. 2018A0303130035.
840
IMDS
119,4
In particular, e-retailers need to ensure timely and efficient delivery to customers in order
to maintain a high quality of service and hence maintain a high degree of customer
satisfaction (Danaher and Mattsson, 1994). Due to the rapid growth of e-commerce, major
e-commerce providers such as Alibaba, Amazon or Zalando have solved technical problems
of button clicksthat deal with millions of simultaneous online orders. However, due to
shortage of labor and equipment, it is difficult for the logistics operations to complete the
large number of e-commerce orders. For instance, on November 11, 2012, Taobao, one of the
largest e-retailer platforms in China, sold more than RMB 19bn (more than 3bn US dollars)
in total sales and exceeded 100m transactions (Zhang et al., 2013). More than 70m packages
were packed and sent to customers, which far exceeds their processing capabilities.
It brought great delays for most online shops and they were complained by customers
(Zhou, 2013). Since delivery requires high investment (Gong et al., 2018), how to maintain
customer delivery satisfaction under limited resources is an ongoing research topic.
Many studies have proposed different optimization methods to strike a balance between
multiple objectives to maintain customer satisfaction in delivery. They proposed various
algorithms, methods and strategies to reduce costs by optimizing delivery schedules and
planning. They studied a list of many critical factors such as delivery time, delivery cost,
capacity, transportation, technology, product delivery service providers, logistic companies,
government regulations, delivery channels, etc. (Zhang, Nemhauser, Sokol, Cheon and Keha,
2018; Zhang, Qin, Yu and Liang, 2018; Liu et al., 2018; Lu et al., 2016; Shankar et al., 2018;
Zarbakhshnia et al., 2018; Shi et al., 2018). In these approaches, they do not consider
individual customer differences. They treat every customer equally. However, customers are
different in the real world. Some of them are more sensitive to delivery service than others.
Therefore, if we can understand the individual customer delivery satisfaction model, we can
improve overall delivery satisfaction by prioritizing customer delivery order without
increasing processing capability.
Recently, the use ofinformatics to extract useful knowledge from e-commerceis powerful
(Wu and Lin, 2018). Thanks to the development of the automation and computing
technologies such as Internet of Things, companies can obtain more information from
customers and they can understand customerspurchasing behaviors and preferences better
than before. In particular, verified consumers write product reviews after purchasing
products. It provides dynamic and trustful feedback data from the customers point of view.
Through the use of machine learning and semantic analysis, many studies have been
conductedto extract useful information fromonline product reviews (e.g. Raviand Ravi, 2015;
Wang, Li, Tian, Liu and Tsui, 2018; Li et al., 2018). They showed that it is possible to model
individual customer delivery satisfaction. In particular, automated machine learning
(AutoML)is the latest technology that makes machinelearning easier to access by automating
the processes of data pre-processing, featureselection, selection of machine learningmethods
and optimization of hyper-parameters (Guyon et al., 2016). They have been successfully
applied to different data mining applications. The results showed that AutoML can compete
with machine learning specialists in many tasks. However, the study of machine learning of
modeling individual customer delivery satisfaction has received less concern.
In this paper, we propose using AutoML to model individual customer delivery
satisfaction to improve overall delivery satisfaction. The main contributions of this paper
include: first, existing research on delivery optimization focuses on the factors such as
delivery cost, delivery time, resources and so on. The concern of individual customer
delivery satisfaction is still limited. This study presents a new perspective with the focus on
modeling individual customer delivery satisfaction to improve overall delivery satisfaction.
Second, the modeling of individual customer delivery satisfaction using machine learning
method is received less concern. An AutoML method is proposed in this study to model
individual customer delivery satisfaction. It helps minimize the knowledge required to train
841
Individual
customer
delivery
satisfaction

To continue reading

Request your trial

VLEX uses login cookies to provide you with a better browsing experience. If you click on 'Accept' or continue browsing this site we consider that you accept our cookie policy. ACCEPT