1. bookAHEAD OF PRINT
Informacje o czasopiśmie
License
Format
Czasopismo
eISSN
2444-8656
Pierwsze wydanie
01 Jan 2016
Częstotliwość wydawania
2 razy w roku
Języki
Angielski
Otwarty dostęp

Research on the construction of early warning model of customer churn on e-commerce platform

Data publikacji: 30 Aug 2022
Tom & Zeszyt: AHEAD OF PRINT
Zakres stron: -
Otrzymano: 12 Apr 2022
Przyjęty: 24 Apr 2022
Informacje o czasopiśmie
License
Format
Czasopismo
eISSN
2444-8656
Pierwsze wydanie
01 Jan 2016
Częstotliwość wydawania
2 razy w roku
Języki
Angielski
Introduction

In recent years, with the rapid development of Internet information technology, we have entered the era of big data. More and more Internet users begin to enjoy the convenient experience brought by online shopping and mobile payment, and they choose to purchase the services or goods they need online [1]. Under such a market background, how to win over high-quality and stable customer base is an important factor to ensure the robust growth of enterprises, at the same time, stable and repeatable customers are the source power for the continuous development of e-commerce enterprises [2, 3]. According to the survey results of China Internet Network Information Development Center [4], nearly 10% of the users in the whole e-commerce market will automatically terminate their contact with one e-commerce platform and switch to another in half a year, which makes the customer mobility of the e-commerce platform appear to be greater. In addition, most customers have less repeat purchase behaviour, which makes the customer churn rate of the e-commerce platform relatively high. In this non-contractual environment, customer relationship is not constrained, which makes it difficult to predict and define the loss.

In recent years, the research on prediction of customer churn through data mining technology has gradually increased [5, 6]. Analysing a series of browsing behaviour data such as search, click, collection and evaluation of users on the e-commerce platform can help enterprises to understand the purchase process, preference and habits of users, as well as the content of interest, etc., and combining with the dimensional subdivision of data can provide more targeted services to customers and provide support for enterprise personalised service strategy, so as to meet the needs of different levels of customers, and improve their satisfaction [7,8]. However, the abovementioned researches mainly focus on the algorithm accuracy or model fusion optimisation, while the research on more comprehensive customer feature description is less. Meanwhile, the definition of customer churn in the model is dualistic, and the description of process dynamics is slightly insufficient [9, 10]. Therefore, by fully considering the characteristics of the e-commerce industry, and combining with the current situation of customer churn of e-commerce enterprises, we can deeply understand the substantive problems behind the customer churn, so as to build an early warning model of customer churn on e-commerce platform. It can effectively predict and judge the possible risk of customer churn for enterprises in the future.

Analysis of customer churn on e-commerce platform

Customer churn in e-commerce enterprises refers to the phenomenon that users temporarily or permanently terminate the transaction relationship with e-commerce enterprises and no longer accept the products or services provided by e-commerce enterprises [11]. As a kind of general customer churn, it is a kind of customer churn under the non-contractual relationship situation.

Definition of customer churn

The loss of customers can be defined as shown in Figure 1.

Fig. 1

Definition of customer churn on e-commerce platform.

Based on result status

When individuals feel or think that the quality or benefits of an organisation are declining, there are three choices: exit, voice and loyalty. Therefore, customer churn is defined as the user's behaviour to terminate the relationship with the enterprise/organisation, which can be summarised as: cancellation, transfer, suspension, etc. If the customer gives up the cooperation with the original enterprise and turns to other enterprises for some reasons, it is generally considered as the loss of the customer.

Based on customer transaction

In the judgement based on the transaction behaviour of customers, customer churn is the process in which the existing customers terminate the activities of enterprise products and services and purchase corresponding products and services from other enterprises or competitors. All the indicators that can describe the customer's trading behaviour are used for the judgement basis. For example, customers in a certain period of time, there will be a larger trading cycle.

Changes in customer churn

In the actual transaction process, there is a kind of customers whose purchase frequency or frequency of products and services provided by the enterprise is only slightly lower than before, but not completely turn to zero. Therefore, customer churn should be classified into two categories: one is based on change of process and the other is based on change of final state [12].

By comparing the above three types, it can be seen that the definition based on the customer result status is a method that directly divides the loss status according to certain business relationship rules, which is easier to identify and more obvious to distinguish customers. But it will make the partition result too one-sided, once the rules are limited, thus the whole prediction judgement will become inflexible; The other method is based on the judgement of the change of customer's trading behaviour, and observes the change in the process of user's transaction behaviour from multiple dimensions, and can trace the cause of the loss based on the final state result.

This paper adopts the definition method based on transaction behaviour; once the value of the observation dimension exceeds or is lower than the preset threshold, then the customer status is determined as churn.

Evaluation method of customer churn
Customer value

In order to more accurately define and describe customers, understand the behaviour characteristics and preferences of customers, and then provide targeted marketing for them. Managers often choose to construct user portrait features and behaviour characteristics, which can maintain long-term interactive relationship with customers according to their own understanding of customers.

As shown in Figure 2, Recurrence, Frequency, Monetary (RFM) model can better measure customer value. The model includes three indicators [13, 14]: R (recency), F (frequency) and M (monetary) reflecting the customer's latest consumption interval. They represent respectively that during the observation period, the time interval between the last consumption behaviour of users and the observation point, the total number of consumption behaviours of users in the observation period, and the total consumption of customers in the observation period.

Fig. 2

Customer value measurement indicators.

Recurrence

It reflects the time interval between the customer's last consumption behaviour and the observation point in the observation period. The larger value of indicator R means that customers have not purchased anything on the platform for a long time, and they are in the decline period of life cycle and who are about to be lost, and the customer value of such customers will be lower. On the contrary, the smaller the R value is, the closer the time interval between the customer's last shopping and the observation point, the higher the customer value.

Frequency

Indicator F is a measure of the number of times customers spend in the inspection period. Generally speaking, the higher the frequency of customers’ consumption, the higher the value. While the shorter the last consumption interval and the higher the consumption frequency in the observation period, the higher the comprehensive value of such customers and the more economic benefits they can bring to enterprises, which should be the focus of the enterprise.

Monetary

This is the most important measurement index of customer value. The higher the consumption amount, the higher the contribution, and the higher the customer value. By synthesising these three indicators, the value of customers to the enterprise can be described in a multi-dimensional way, and according to the different combination performances of different customer groups with the three indicators, some of their behaviour characteristics can be analysed.

Life cycle customer

Customer's life cycle describes the transformation process of customer's state from establishing contact to ending contact between customer and enterprise, and reflects customer's characteristic performance in each stage of enterprise. In order to explore the key to customer churn, many researchers are committed to the study of customer change characteristics in various stages of the life cycle [15, 16]. As shown in Figure 3, the whole life cycle is divided into five stages as follows:

Fig. 3

Life cycle of customer.

Acquisition

In this period, enterprises explore potential customers from the market by means of marketing, and customers also understand the actual situation of enterprises through various kinds of information, so as to preliminarily understand each other and lay the foundation for the development of follow-up relations.

Promotion

In this period, with each product or service linkage between the two sides, the satisfaction and cooperation intention between the two sides will accelerate development, and the interdependence relationship will also deepen.

Mature

Due to the interaction in the first two stages, enterprises and customers will have more and more understanding in this period, so the relationship between them is further improved where all kinds of contact and cooperation relations are gradually become stable and for enterprises, this is the golden period with customers.

Recession

Due to poor communication, or competition from other companies, the established good customer relationship before has deteriorated gradually and the customers showed various negative performances, and the transactions gradually slowed down. However, it should be realised that the deterioration of the relationship may occur at any stage of the whole customer cycle, which is not limited to the last stage of the life cycle.

Loss

After the recession stage, customers lose their trust on the enterprise at last, which make them to end the business contact with the enterprise. At this stage, the customers are completely lost.

With the theoretical study of the life cycle, we can intuitively understand the stage of customer churn, and help enterprises to develop better customer retention strategies after the subsequent analysis of customer churn.

Early warning model of customer churn risk on e-commerce platform
Function of early warning of customer churn

The early warning of customer churn management occurs before the real loss of customers, which can be sensitive to the impact of customer churn dynamic factors; once there is movement, it will send a signal to remind enterprises to take necessary measures to improve the status quo. The early warning management of customer churn can not only help enterprises to achieve their goals, but also prevent them from falling into severe difficulties, which is important to improve enterprise's own value and realise excellent management. As shown in Figure 4, it usually includes the following aspects [17, 18]:

Fig. 4

Function of early warning of customer churn.

Monitoring the signs of customer churn

Detecting the signs of customer churn needs to be carried out under a comprehensive plan. It is the first step for enterprises to predict customer churn in order to reduce the number of customers churn. The main job of detecting the signs of customer churn is to detect the customers who are about to be lost in real time.

Judgement of customer churn

Judgement of customer churn, that is, after detecting the signs of customer churn, scientific methods are used to get the index data for in-depth analysis of the index data which exposed customer behaviour characteristics, thus helps to judge the tendency of customer churn.

Assessing customer churn

Evaluation of customer churn should be carried out after the completion of the previous links. After the data of customer churn sign detection link is saved, these index data are analysed in depth, and the abnormal indicators are scored and summarised, and then graded according to different severity. Finally, the enterprise formulates a reasonable weight determination standard according to the actual situation; comprehensive evaluation of the customer churn status of enterprises is made by comprehensive factors.

The concept of customer churn warning determines that the system can be used to monitor customers with significant churn tendency. However, in order to effectively solve the problem of customer churn, it is necessary to construct a comprehensive, scientific and reasonable customer churn early warning model on the premise of deeply exploring the early warning mechanism and mode of customer churn.

Classification of customer churn risk levels

Cluster analysis is a multivariate statistical method, which refers to grouping abstract objects into sets of similar elements. It is usually used to study the classification of samples or indicators [19]. The essence of clustering analysis is to reasonably and fully replace unknown information with known information, classify and identify the essential attributes of grey system, and obtain objective quantitative analysis results.

The sample data is clustered by K-means, and 100 samples are clustered into four types of churn risk states, namely: safe (A), less safe (B), dangerous (C) and very dangerous (D). The cluster analysis and evaluation description of each risk state are shown in Table 1.

Classification of customer churn risk levels.

Proportion Customer description
Safety (A) 14% Most of these users think that the platform has fast response speed, beautiful interface, convenient use, good customer service, payment method that can meet the basic needs of users, fast delivery speed, satisfaction with the service of delivery staff.
Less safe (B) 35% However, such users sometimes give up payment because they can’t find a suitable way, and worry about whether their payment security and personal privacy will be leaked from time to time. They have a high degree of participation in product evaluation, high requirements for products and will return and exchange products whose descriptions don’t match the real ones for a long time.
Danger (C) 20% These users usually feel that the response speed of the platform is slow, their satisfaction with the products recommended by the e-commerce platform is low and their delivery speed and delivery staff service are often not satisfied. There is no dependence on the exclusive sales of products, and they think that the prices of most products are somewhat high, and the participation in the evaluation of purchased products is also low.
Very dangerous (D) 31% Their satisfaction with the platform is very low, which is usually reflected in the dissatisfaction with the aesthetics and convenience of the interface, the quality of the goods and the service of the delivery staff.
Model construction
Index selection

The e-commerce transaction is divided into several links, such as registration and login, interface browsing, payment, commodity purchase and after-sales service. Therefore, on the basis of previous research literature, considering the obvious characteristics of e-commerce enterprises and combining the actual situation of e-commerce enterprises in China, the index system shown in Figure 5 is constructed.

Fig. 5

Construction of early warning model index.

Structural design

In this paper, the early warning model based on neural network is adopted, and the three-layer network topology is selected according to the data of training sample and prediction sample, as well as indicators of early warning mentioned earlier. The network structure is shown in Figure 6.

Fig. 6

Network structure of early warning model.

Input layer

It is located at the bottom of the neural network, where neurons in this layer receive pre-selected indicators as input information from the external environment. In the neural network model of this paper, the number of neurons in the input layer is determined by the input amount of index data and 15 normalised index data of early warning of customer churn are respectively input as input samples, which are used to predict the crisis situation of customer churn.

(2) The hidden layer, in which the hidden layer nodes find out the hidden rules from the sample data, and then store the information in the form of connection weights. In summary, the following four formulas can be used to calculate hidden nodes:

L=log2m L = {\rm{log}}_2^m L=(m×n)/2 L = (m \times n)/2 L=m+0.618(mn) L = m{\kern 1pt} + {\kern 1pt} 0.618(m - n) L=2m+1 L = 2m + 1

Among them m is input layer node, n is the output layer node.

In this paper, the four empirical formulas are used to calculate the number of nodes in the four hidden layers, and the different nodes are put into the neural network for training. Then, the sample data of test set is brought into the neural network for fitting and the prediction accuracy of the output layer data obtained is compared with the expected result of the test set sample. The number of hidden layer nodes with the highest prediction accuracy 8 is selected as the number of hidden layer nodes in the early warning model.

Output layer

It is located at the highest level of the network. At this level, the neural network transmits the hidden layer data to the output neuron, which in turn transmits the information to the external environment. The data category of the output layer determines the number of neurons in the output layer.

In this paper, when the network output layer is defined, the alert levels are determined in four levels according to different levels of customer churn risk. The result of the first-level warning level corresponding to the output layer is [0001], the second-level warning level corresponding to the output layer is [1000], the third-level warning level corresponding to the output layer is [0100] and the fourth-level warning level corresponding to the output layer is [0010].

Selection of functions

The customer churn warning system of e-commerce platform based on neural network mainly processes data through transfer function and training function, and tests the training effect of the model by learning rate and expected error.

Transfer function

Transfer function is usually selected from purelin, tansig and logsig. As there are negative values in the input data, in order to prevent data compression from affecting the accuracy, the tansig function that can map the neuron input range from (−∞, +∞) to (−1,1) is selected as the transfer function of hidden layer, and its expression is:

tansig(m)=2(1+e2m)1 {\rm{tan}}{\kern 1pt} {\rm{sig}}(m) = {2 \over {\left( {1 + {e^{ - 2m}}} \right) - 1}}

In addition, as the expected output data of the output layer are [1000], [0100], [0010] and [0001], this paper determines that the transfer function from the hidden layer to the output layer is the logsig function, which maps the input range of neurons from (−∞, +∞) to (0, 1), and its function is logsig(m)=1(1+em) {\rm{log}}{\kern 1pt} {\rm{sig}}(m) = {1 \over {\left( {1 + {e^{ - m}}} \right)}}

Training function

Trainlm is selected as the training function in this paper. When the number of network weights is small, it can quickly converge, and has strong advantages in improving training speed, which avoids falling into local minimum. It is suitable for the function fitting of small and medium-sized networks, with fast convergence and small error.

Learning rate

In order to ensure the stability of the network, we usually try to choose a small learning rate while ensuring the training duration. The most scientific and reasonable method is to first determine the learning rate of several different values, and then, after the training of neural network, select the most appropriate learning rate according to the descending speed of the error function. Generally, the range of selection is between 0.01 and 0.8. Through the test, the learning rate of 0.01 is chosen.

Expected error

Usually, the smaller the expected error value is, the more accurate the model. However, if the expected error value is too small, the number of hidden layer nodes will be increased unintentionally, resulting in the extension of the training time of neural network. Meanwhile, it will greatly increase the risk of ‘over-fitting’ of the network in training, which directly leads to the reduction of the universality of the neural network. In this paper, the following indicators are used for calculation:

Mean absolute error MAE=|ej|N MAE = {{\sum \left| {{e_j}} \right|} \over N}

Sum square error SSE=(ej)2 SSE = \sum {\left( {{e_j}} \right)^2}

Mean square error MSE=(ej)2N MSE = {{\sum {{\left( {{e_j}} \right)}^2}} \over N} where ej represents prediction error (ej=yjkojk) \left( {{e_j} = y_j^k - o_j^k} \right) , yj represents the predicted value of output and oj represents the actual value.

Generally, the expected error is between 0.0001 and 0.01. The expected error is set to 0.001 with the help the test.

Model test

After the neural network model with good fitting degree is obtained, it is necessary to test the early warning accuracy of the established, so as to verify whether the prediction accuracy of the model meets the requirements. First, the early warning index data of customer churn risk of sample users collected in the above survey is used as input layer data and input into the neural network, and after the processing of the hidden layer network, the output layer data is obtained. By comparing the output results obtained by hidden layer processing with the actual output results, the smaller the error value, the higher the fitting degree of the early warning model based on neural network to the variables of output. In general, the high fitting degree represents the high applicability and generalisation ability of the early warning model.

By further transforming the output results of neural network, the risk warning grades of ten test set samples from the model constructed can be obtained. After transformation, we can get the output results of test set samples, and the comparison between the customer churn risk warning grades and the expected output is shown in Table 2.

Test results of early warning model.

Actual output Expected output Prediction result Warning level
(0, 0, 0, 1) (0, 0, 0, 1) Correct A
(0, 0, 0, 1) (0, 0, 0, 1) Correct A
(0, 1, 0, 0) (0, 1, 0, 0) Correct C
(0, 0, 1, 0) (0, 0, 1, 0) Correct D
(0, 0, 1, 0) (0, 0, 1, 0) Correct D
(0, 1, 0, 0) (0, 1, 0, 0) Correct C
(0, 0, 1, 0) (0, 0, 1, 0) Correct D
(0, 1, 0, 0) (0, 1, 0, 0) Correct C

By comparing the expected output of the test set sample with the output of the early warning model based on neural network, it can be seen that the forecast of the risk warning is in the sequence of first-level warning, first-level warning, third-level warning, fourth-level warning, fourth-level warning, third-level warning, fourth-level warning and third-level warning. The result of actual output is completely consistent with the expected output and so the accuracy rate has reached 100%, which means that the prediction data obtained by this early warning model is not far from the real data, and the prediction accuracy rate of it is high.

Conclusion

Based on the phenomenon of customer churn in e-commerce enterprises, this study puts forward the early warning of risk for enterprises. By summarising the related research results of customer churn theory and risk early warning of enterprise, this paper extracts 15 risk early warning indicators about customer churn of e-commerce, constructs the risk early warning model of e-commerce customer churn by using neural network, and tests this model. The results show that the early warning index system of customer churn risk in China's e-commerce enterprises designed in this paper is scientific and reasonable. The early warning system of customer churn risk of e-commerce platform based on neural network has excellent generality, and its prediction accuracy is 100%, which can provide important reference value for the early warning research of customer churn risk in the whole e-commerce industry.

Fig. 1

Definition of customer churn on e-commerce platform.
Definition of customer churn on e-commerce platform.

Fig. 2

Customer value measurement indicators.
Customer value measurement indicators.

Fig. 3

Life cycle of customer.
Life cycle of customer.

Fig. 4

Function of early warning of customer churn.
Function of early warning of customer churn.

Fig. 5

Construction of early warning model index.
Construction of early warning model index.

Fig. 6

Network structure of early warning model.
Network structure of early warning model.

Classification of customer churn risk levels.

Proportion Customer description
Safety (A) 14% Most of these users think that the platform has fast response speed, beautiful interface, convenient use, good customer service, payment method that can meet the basic needs of users, fast delivery speed, satisfaction with the service of delivery staff.
Less safe (B) 35% However, such users sometimes give up payment because they can’t find a suitable way, and worry about whether their payment security and personal privacy will be leaked from time to time. They have a high degree of participation in product evaluation, high requirements for products and will return and exchange products whose descriptions don’t match the real ones for a long time.
Danger (C) 20% These users usually feel that the response speed of the platform is slow, their satisfaction with the products recommended by the e-commerce platform is low and their delivery speed and delivery staff service are often not satisfied. There is no dependence on the exclusive sales of products, and they think that the prices of most products are somewhat high, and the participation in the evaluation of purchased products is also low.
Very dangerous (D) 31% Their satisfaction with the platform is very low, which is usually reflected in the dissatisfaction with the aesthetics and convenience of the interface, the quality of the goods and the service of the delivery staff.

Test results of early warning model.

Actual output Expected output Prediction result Warning level
(0, 0, 0, 1) (0, 0, 0, 1) Correct A
(0, 0, 0, 1) (0, 0, 0, 1) Correct A
(0, 1, 0, 0) (0, 1, 0, 0) Correct C
(0, 0, 1, 0) (0, 0, 1, 0) Correct D
(0, 0, 1, 0) (0, 0, 1, 0) Correct D
(0, 1, 0, 0) (0, 1, 0, 0) Correct C
(0, 0, 1, 0) (0, 0, 1, 0) Correct D
(0, 1, 0, 0) (0, 1, 0, 0) Correct C

Yu Xiaobing, Cao Jie, Zhang Mengnan. Research on the cause evaluation of B2C e-commerce customer churn [J]. Fuzzy Systems and Mathematics, 2012, 12:167–171 (in Chinese). YuXiaobing CaoJie ZhangMengnan Research on the cause evaluation of B2C e-commerce customer churn [J] Fuzzy Systems and Mathematics 2012 12 167 171 (in Chinese). Search in Google Scholar

Zhang Cuiping. How to strengthen enterprise management to reduce customer churn modern economic information, Journal of Enterprise Reform and Management, 2016, 7:80–81 (in Chinese). ZhangCuiping How to strengthen enterprise management to reduce customer churn modern economic information Journal of Enterprise Reform and Management 2016 7 80 81 (in Chinese). Search in Google Scholar

Tzeng G H, Lin C S. Combined rough set theory and flow network graph to predict customer churn in credit card accounts [J]. Expert Systems with Applications, 2010, 38(1):8–15. TzengG H LinC S Combined rough set theory and flow network graph to predict customer churn in credit card accounts [J] Expert Systems with Applications 2010 38 1 8 15 10.1016/j.eswa.2010.05.039 Search in Google Scholar

Ngai EW T, Li Xiu. Customer churn prediction using improved balanced random forests [J]. Expert Systems with Applications, 2011, 36(3):5445–5449. NgaiEW T LiXiu Customer churn prediction using improved balanced random forests [J] Expert Systems with Applications 2011 36 3 5445 5449 10.1016/j.eswa.2008.06.121 Search in Google Scholar

Zhang Weidong, Li Yuan. Early warning analysis of telecom user churn based on algorithm [J]. Journal of Yibin University, 2014, 14(6):119–122 (in Chinese). ZhangWeidong LiYuan Early warning analysis of telecom user churn based on algorithm [J] Journal of Yibin University 2014 14 6 119 122 (in Chinese). Search in Google Scholar

Fu Jie, Fang Fang, Yan Kewen. Research on prediction and retention of customer churn in communication industry based on logistic regression [J]. Journal of Ezhou University, 2015, 2(6):110–112 (in Chinese). FuJie FangFang YanKewen Research on prediction and retention of customer churn in communication industry based on logistic regression [J] Journal of Ezhou University 2015 2 6 110 112 (in Chinese). Search in Google Scholar

Zhang Qiuju, Zhu Bangzhu. E-commerce customer churn prediction model based on self-organizing data mining [J]. Business Strategy, 2011, 01:95–99 (in Chinese). ZhangQiuju ZhuBangzhu E-commerce customer churn prediction model based on self-organizing data mining [J] Business Strategy 2011 01 95 99 (in Chinese). Search in Google Scholar

Kuanchin Chen, Ya-Han Hu, Yi-Cheng Hsieh. Predicting customer churn from valuable B2B customers in the logistics industry: a case study [J]. Information Systems and e-Business Management, 2014, 10:476–494. ChenKuanchin HuYa-Han HsiehYi-Cheng Predicting customer churn from valuable B2B customers in the logistics industry: a case study [J] Information Systems and e-Business Management 2014 10 476 494 10.1007/s10257-014-0264-1 Search in Google Scholar

Ren Jianfeng, Zhang Xinxiang. Modeling and prediction of customer churn in e-commerce [J]. Computer Simulation, 2012, 5:363–366 (in Chinese). RenJianfeng ZhangXinxiang Modeling and prediction of customer churn in e-commerce [J] Computer Simulation 2012 5 363 366 (in Chinese). Search in Google Scholar

Hadden J. Linking service quality, customer satisfaction and behavioral intention [J]. Journal of Health Care Marketing, 2007(9):5–17. HaddenJ Linking service quality, customer satisfaction and behavioral intention [J]. Journal of Health Care Marketing 2007 9 5 17 Search in Google Scholar

Yu Xiaobing, Wang Xuming. Analysis of the factors affecting e-commerce customer churn [J]. Practice and Understanding of Mathematics, 2014, 11:1–5 (in Chinese). YuXiaobing WangXuming Analysis of the factors affecting e-commerce customer churn [J] Practice and Understanding of Mathematics 2014 11 1 5 (in Chinese). Search in Google Scholar

Ren Jianfeng, Zhang Xinxiang. Modeling and forecasting of customer churn in e-commerce [J]. Computer Simulation. 2012, 5:363–366 (in Chinese). RenJianfeng ZhangXinxiang Modeling and forecasting of customer churn in e-commerce [J] Computer Simulation 2012 5 363 366 (in Chinese). Search in Google Scholar

Dai Yisheng, Shen Peilan, Sun Hongxia. Research on customer churn prediction of e-commerce websites based on Pareto/NBD model [J]. Science, Technology and Engineering, 2010, 9:6792–6795 (in Chinese). DaiYisheng ShenPeilan SunHongxia Research on customer churn prediction of e-commerce websites based on Pareto/NBD model [J] Science, Technology and Engineering 2010 9 6792 6795 (in Chinese). Search in Google Scholar

Hwang H, Jung T, Suh E. An LTV model and customer segmentation based on customer value: a case study on the wireless telecommunication industry [J]. Expert Systems with Applications, 2004, 26(2):181–188. HwangH JungT SuhE An LTV model and customer segmentation based on customer value: a case study on the wireless telecommunication industry [J] Expert Systems with Applications 2004 26 2 181 188 10.1016/S0957-4174(03)00133-7 Search in Google Scholar

Chen Mingliang. Customer loyalty and customer relationship life cycle [J]. Journal of Management Engineering, 2003, 2:90–93 (in Chinese). ChenMingliang Customer loyalty and customer relationship life cycle [J] Journal of Management Engineering 2003 2 90 93 (in Chinese). Search in Google Scholar

Glady N, Baesens B, Croux C. Modeling churn using customer lifetime value [J]. European Journal of Operational Research, 2009, 197(1):402–411. GladyN BaesensB CrouxC Modeling churn using customer lifetime value [J] European Journal of Operational Research 2009 197 1 402 411 10.1016/j.ejor.2008.06.027 Search in Google Scholar

Wang Zhijun. Research on early warning of customer churn based on neural network [D]. Changchun: Jilin University, January 2013 (in Chinese). WangZhijun Research on early warning of customer churn based on neural network [D]. Changchun Jilin University January 2013 (in Chinese). Search in Google Scholar

Manoj Kuma. Trust and e-commerce: a study of consumer perceptions [J]. Electronic Commerce Research and Applications, 2009(2):203–215. Manoj Kuma Trust and e-commerce: a study of consumer perceptions [J] Electronic Commerce Research and Applications 2009 2 203 215 10.1016/S1567-4223(03)00024-3 Search in Google Scholar

Chen Mengyin. Comprehensive evaluation of quality based on principal component analysis and cluster analysis [J]. Food Science. 2016, 9:72–75 (in Chinese). ChenMengyin Comprehensive evaluation of quality based on principal component analysis and cluster analysis [J] Food Science 2016 9 72 75 (in Chinese). Search in Google Scholar

Polecane artykuły z Trend MD

Zaplanuj zdalną konferencję ze Sciendo