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 loyalty prediction of e-commerce customer based on data mining

Data publikacji: 05 Sep 2022
Tom & Zeszyt: AHEAD OF PRINT
Zakres stron: -
Otrzymano: 12 Apr 2022
Przyjęty: 24 May 2022
Informacje o czasopiśmie
License
Format
Czasopismo
eISSN
2444-8656
Pierwsze wydanie
01 Jan 2016
Częstotliwość wydawania
2 razy w roku
Języki
Angielski
Introduction

With the advancement of market economy, entrepreneurs have gradually found that customer resources play an increasingly important role in market competition. Therefore, the ‘customer-centred’ marketing model has gradually replaced the traditional ‘product-centred’ marketing model, and the research on customer resources has become a hot spot of common concern in theoretical and business circles [1, 2]. To maintain its market position and sustainable competitive advantage, the prediction of customer's consumption behaviour is very important. The fundamental purpose of studying the data of daily consumption behaviour among customers is to identify the most valuable customers of the enterprise, and to market their products, so as to increase their profitability. Studies have shown that [3] 80% of an enterprise's total marketing profit comes from enterprise marketing and 20% comes from the contribution of loyal consumers. In addition, some studies have shown that [4] the transaction cost between enterprises and regular customers is only 1/6 of that between them and the total customers. In the existing e-commerce market, apart from the fierce competition between the original e-commerce platforms, in recent years, with the change of social mode, a variety of emerging e-commerce modes are added such as WeChat, Weibo, community, etc., which also makes the e-commerce market more competitive [5, 6].

Customer loyalty is analysed according to the store situation, user psychology and user behaviour data, and the method of predicting repeated purchase behaviour is designed to reflect the importance and attention of customer loyalty [7]. Theoretically, customer loyalty can be discussed from the data generated by consumers’ purchasing behaviour, which can also be discussed from the factors that influence repeated purchase behaviour. However, in the field of academic research, scholars tend to theoretically discuss the factors that influence customer loyalty, thus neglecting the research on behaviour data [8]. At the same time, in the process of studying consumers’ consumption intention and purchasing behaviour, the differences between consumers will be ignored, and the conclusion is usually unreasonable, which leads to insufficient precision of indicators such as accuracy and recall rate [9]. Therefore, it has become an urgent problem to study how to improve the prediction of consumer loyalty.

Churn and loyalty of customer
E-commerce customer churn

Customer churn is the basis for effectively judging customer loyalty [10]. In this paper, the concept of e-commerce enterprise customers’ loss is defined as: if e-commerce customers’ consumption behaviour occurs again within a set period of time after consuming on the platform, it is regarded that the customer has not lost, otherwise, if there is no consumption behaviour within a set period of time, it is determined that they have lost. However, the consumption of customers in e-commerce enterprises generally changes periodically, so customers who have no consumption behaviour in a certain period can’t simply and directly judge that the customers are lost. The reasons for the churn of e-commerce customers are shown in Figure 1.

Fig. 1

Factors of e-commerce customer churn.

E-commerce customer loyalty

E-commerce customer loyalty can be defined as online consumers’ loyalty to products and services provided by enterprises on the Internet based on past shopping experiences. It refers to the loyalty of online customers to an online store or a brand of products or services on the e-commerce platform [11, 12].

Behaviour and intention of repeated purchase among consumers are the basis of loyalty classification of e-commerce customer. As shown in Figure 2, it can be divided into the following four types: persistent loyalty, potential loyalty, false loyalty and disloyalty.

Fig. 2

Classification of e-commerce customer loyalty.

In this paper, the quantitative index used to measure the loyalty of e-commerce customers is the probability of repeated purchase of a certain merchant on the e-commerce platform, which is divided into the following three parts: >70% are highly loyal, >70–30% are moderately loyal and <30% are low loyal that needs to be maintained by merchants.

Application basis of data mining in loyalty prediction

Data mining refers to finding out hidden rules and relations from a large amount of data information, which means relying on large-scale original data and pattern analysis of data mining to obtain corresponding knowledge. Based on the perspective of business, data mining needs to extract the data from the huge database, and carry out the corresponding elaboration and analysis through specific rules, so as to achieve the desired modelling effect [13, 14].

Application scene

The research of data mining technology in prediction of e-commerce customer loyalty mainly includes five tasks as shown in Figure 3.

Classification analysis.

Classification refers to the analysis of data objects with known categories, and then the search for other data by analysing the data and distinguishing their categories or detailed models, which can predict the specific types of unknown objects. By using the matching scheme of data mining in the e-commerce environment, the purchasing activities of multiple age groups and genders can be described effectively. On the basis of classification, it can be referred to the similar characteristics of customers of the same kind and then formulate corresponding marketing measures to provide targetted information services [15].

Prediction and evaluation.

Carry out in-depth collation of the actual data, and further carry out follow-up comprehensive research to ensure the timeliness and regularity of the final model calculation, so as to predict and evaluate the development trend and results of future events [16].

Discover time series patterns.

Through pattern discovery, other similar time series events can be found, and the corresponding patterns with high probability of recurrence can be further searched. The model helps e-commerce to better predict basic user activities, detect the sequence of user preferences, and further provide more personalised service mechanism.

Cluster analysis.

The main difference between cluster analysis and classification analysis is that data objects do not have corresponding class labels, and cluster analysis can further divide data formation into classes or clusters by referring to specific relationships or attributes, so that data objects in the same class or cluster are highly similar, while data objects in different classes or clusters are quite different from each other.

Association rules.

Association rule refers to finding frequently occurring items from a large amount of historical data, and finding the hidden association data among the data. The specific rules in the e-commerce model are to further explore the detailed number of clicks and the relationship between collection and purchase with reference to the sequential exploration activities of customers [17].

Fig. 3

Tasks of data mining technology in prediction of e-commerce customer loyalty.

Application process

The data mining process of e-commerce platform can be divided into six steps, as shown in Figure 4.

Identify business.

In the data mining carried out by e-commerce, the first procedure is to judge detailed business problems and confirm the core purpose of subsequent data mining, which is difficult to predict its results effectively in real life.

Collect and select data.

The second step in the data mining process of e-commerce platform is to collect and select data, which is to extract information more in line with the goal of data mining through specific data, appropriate data can also reduce the workload of data mining and improve its quality [18].

Data preprocessing.

The third step in the data mining process of e-commerce platform is the basic pretreatment. Pre-processing can pre-process specific data sets, deal with missing data and filter abnormal data, and build feature information as the basis of later model building, which is helpful to improve the accuracy of data mining model.

Build a data mining model.

Refer to the first step to confirm the detailed goal, further select the mining algorithm that meets the requirements, and form the algorithm model that meets the expectations.

Data mining.

Relying on the model built by the previous programme, further research on the basic data of the third step is carried out to obtain ideal information and data.

Analyse and evaluate the results.

This process ensures that the results obtained after mining can be clearly and intuitively displayed after eliminating irrelevant and redundant patterns.

Fig. 4

Data mining process of e-commerce platform.

E-commerce customer loyalty prediction model based on data mining
Algorithm selection
XGBoost algorithm flow

Although there are many methods to predict the customer loyalty, the existing prediction and classification methods have certain limitations to predict the purchase possibility of customers when they hold activities in a shop because the data of different industries and businesses have different characteristics. As the data of customers’ purchasing behaviour on e-commerce platform has the characteristics of limited dimension, abundant data and discrete type, the effective features of data are category data and the logs generated by the past historical records of customers’ purchasing corresponding products in a store on Tmall platform. Therefore, in this study, the classifier model can’t get the ideal effect. So the Boosting algorithm is selected in machine learning to estimate customer loyalty.

XGBoost is a GBDT algorithm and its corresponding generalised algorithm in the Gradient Boosting architecture. Generally, the objective function and the prediction function will be constructed. By applying the training sample to the lowest objective function and making it learn the relevant parameters, detailed samples can be further classified and predicted. Equation (1) shows the specific form. Obj(θ)=L(θ)+Ω(θ) Obj\;(\theta ) = L(\theta ) + \Omega (\theta ) where Ω(θ) represents a punishment for errors in complex models and L(θ) is an error function, which is used to evaluate the fitting degree of data and model which is defined as Eq. (2): yi=K1Kfk(xi),fkF {y_i} = \sum\limits_{K - 1}^K {f_k}({x_i}),{\kern 1pt} {f_k} \in F

Specific function f regard as the basic function of F, while F is regarded as a special set of all regression trees, as shown in Eq. (3). Obj(θ)=inl(yi,yl)+k=1KΩ(fk) Obj\;(\theta ) = \sum\limits_i^n l({y_i},{y_l}) + \sum\limits_{k = 1}^K \Omega ({f_k})

The basic principle of Boosting is to ensure that a specific model remains fixed, and further add new functions f, so as to control the detailed objective function as much as possible. Obj(t)=inl(yi,yl(t1)+ft(xi))+Ω(fi) Ob{j^{(t)}} = \sum\limits_i^n l\left( {{y_i},y_l^{(t - 1)} + {f_t}({x_i})} \right) + \Omega ({f_i})

Use Taylor formula to expand and define the objective function approximately, as shown in Eq. (5). Obj(t)i=1n[l(yi,yl(t1))+gift(xi)+12hift2(xi)]+Ω(ft) Ob{j^{(t)}} \approx \sum\limits_{i = 1}^n \left[ {l\left( {{y_i},y_l^{(t - 1)}} \right) + {g_i}{f_t}({x_i}) + {1 \over 2}{h_i}f_t^2({x_i})} \right] + \Omega ({f_t})

When the constant term is removed, the objective function Obj(θ) is relatively dependent on the error function L(θ), as shown in Eq. (6). Obj(t)i=1n[gift(xi)+12hift2(xi)]+Ω(ft) Ob{j^{(t)}} \approx \sum\limits_{i = 1}^n \left[ {{g_i}{f_t}({x_i}) + {1 \over 2}{h_i}f_t^2({x_i})} \right] + \Omega ({f_t})

Then the specific tree is further divided into a leaf weight W and a structure part q, W will give the leaf score corresponding to each index number: ft(x) = wq(x), where wRT,q:Rd{1,2,3,4,,T} w \in {R^T},{\kern 1pt} q:{R^d} \to \{ 1,2,3,4, \cdots ,T\}

Use Eq. (5) to judge the particularity of the decision tree, and the complexity also covers the detailed number and detailed score L2 and its square of the module, the specific formula is shown in the following Eq. (8): Ω(ft)=γT+12λj=1Twj2 \Omega ({f_t}) = \gamma T + {1 \over 2}\lambda \sum\limits_{j = 1}^T w_j^2 where wj2 w_j^2 is the magnitude of w and L2 squared, and T is the number of leaves in the tree. In an objective function L(θ), T uncorrelated quadratic functions of one variable are set, which are defined as shown in Eq. (9): Gj=iIjgi,Hj=iIjhi {G_j} = \sum\limits_{i \in {I_j}} {g_i},{H_j} = \sum\limits_{i \in {I_j}} {h_i}

Bring Eq. (10) into the above formula to obtain: Obj(t)=j=1T[Gjwj+12(Hj+λ)wj2]+γT Ob{j^{(t)}} = \sum\limits_{j = 1}^T \left[ {{G_j}{w_j} + {1 \over 2}({H_j} + \lambda )w_j^2} \right] + \gamma T

The derivative of wj is obtained by Eq. (11), whose result is 0. The optimal solution of W and its corresponding objective function L(θ) are solved, as shown in Eq. (12): wj*=GjHj+λ w_j^* = - {{{G_j}} \over {{H_j} + \lambda }} Obj=12j=1TGj2Hj+λ+γT Obj = - {1 \over 2}\sum\limits_{j = 1}^T {{G_j^2} \over {{H_j} + \lambda }} + \gamma T

Algorithm optimisation

The research and analysis of customer churn is a classic two-category problem. Due to the deviation of the prediction model itself, the positive and negative cases are often wrongly divided, and the customers who have lost are mistaken for those who have not lost. On the contrary, the customers with higher loyalty are mistaken for those who have lost, which brings a lot of inconvenience to the daily management of enterprises, and even wastes the limited resources of enterprises. Therefore, the definition of loss function of XGBoost algorithm is optimised in this paper to reduce the loss caused by wrong classification.

Logarithmic function is a loss function commonly used in solving classification problems, as shown in Eq. (13): L(θ)=i[yiln(1+ey^t)+(1yi)ln(1+ey^i)] L(\theta ) = \sum\limits_i [{y_i}\ln \;(1 + {e^{ - {{\hat y}_t}}}) + (1 - {y_i})\ln \;(1 + {e^{{{\hat y}_i}}})]

In the calculation process of XGboost algorithm, it is easy to find out through the loss function, which sets the probability of positive and negative examples being divided into the same. But in fact, there are more cases to divide positive examples into errors. So penalty coefficient α is added to the loss function to distinguish the probability of two different errors.

The definition of the loss function after adding the penalty coefficient is shown in Eq. (14), because there are more cases in which the positive examples are wrongly divided. The value range of α is (0.5, 1]. L(θ)=i[yiln(1+eαY^i)+(1yi)ln(1+e(1α)y^i)] L(\theta ) = \sum\limits_i [{y_i}\ln \;(1 + {e^{ - \alpha {{\hat Y}_i}}}) + (1 - {y_i})\ln \;(1 + {e^{(1 - \alpha ){{\hat y}_i}}})]

Data description

This data set mainly describes the sales to customers in Tmall platform and the purchase records of customers in the past 6 months, including the long-term user behaviour logs of customers on Tmall platform. According to the attributes of different data, it can be divided into user information, commodity information and Scenario information. User information includes user ID, province and city; commodity information includes user logo, commodity name, price, category and shelf time; scenario information includes payment time, order quantity and order status.

Data processing

Before the data is imported into R, it is preliminarily sorted out. The steps are divided into four steps, as shown in Figure 5.

Step 1: Use SQL Server software to extract customers’ purchase behaviour from file of text format.

Step 2: Adopt the ifelse() function in R and as the date function to convert it into the format of time data and specifies the format as yyyy-mm-dd.

Step 3: As the independent variables corresponding to individual dependent variables are all empty, delete them directly.

Step 4: Rename the original variables and interpret the variables in the dataset.

Fig. 5

Processing steps of e-commerce customer data.

Data cleaning

Since the data are made ‘dirty’ by missing data, abnormal data and repeated data, we need to clean it.

Process of missing values

In view of the actual gender deficiency, the specific maximum possible number was used to carry out the process of follow-up filling. So that the effective statistics of all genders can be carried out and the maximum value can be filled effectively. In view of the deficiency of the age range, the subsequent filling operation can be carried out through the specific mean value, so that the specific age can be calculated effectively. The corresponding mean value data are obtained to achieve the ideal filling effect.

Outlier handling

Local outlier factor combines anomaly detection and density-based clustering algorithm where any natural number k is set and the distance between p and some object O as the k-distance of p are defined. While the k-distance neighbourhood of P contains all objects that are within k-distance of P. The principle of calculating local anomaly factor is as follows:

The calculation of local density of object p LD(p)=|N(p)|/oN(p)Dist(p,o) LD(p) = |N(p)|/\sum\limits_{o \in N(p)} {\rm{Dist}}(p,o)

Among them, Dist(p, o) is the n-dimensional Euclidean space distance between point p and o. |N(p)| is a field of his neighbour's number of the point p.

Local deviation index: If the object p neighbourhood is not dense, namely, |N(p)| < MinPts, the partial deviation index of p is defined as: LDI(p)=oN(p)(LD(o)/LD(p))|N(p)| LDI(p) = {{\sum\nolimits_{o \in N(p)} {(LD(o)/LD(p))} } \over {|N(p)|}}

The local deviation index of object P represents the degree to which object P deviates from its neighbours.

In this paper, LOF algorithm is used to eliminate the outliers in the preliminary data set. The theoretical basis is to analyse the local density of a point, in order to compare it with the density of points distributed around it. If the comparison shows that the density of the area where this point is located is sparse, it indicates that this point is an abnormal value. In addition, the outlier score is calculated by the lofactor function in DMwR package, and the data with the top five scores are taken as outliers, and then eliminated. In this paper, a total of 200 pieces of abnormal data were eliminated.

Implementation process

Based on the above analysis, in this paper, the xgb.train () function in xgboost package in R language is adopted, and the construction process of e-commerce customer loyalty prediction model proposed in this paper is as follows:

Import data into R, and clean the data.

Store the cleaned data in R software in the format of dataframe, and divide it into training set (70%) and testing set (30%).

Determine that the data in the training set and the test set have the same distribution, and test the homogeneity of the samples by ks.test () function.

Set the parameter variables in xgb.train () function, and obtain different prediction results through different parameter settings, thus select the best prediction result.

Substitute the test set into the model, and use the prediction function to evaluate it.

Establish the confusion matrix, so that the gap between the predicted results and the real values can be obtained to judge the accuracy of the model prediction.

Evaluation and comparison of e-commerce customer loyalty prediction models.

Accuracy Precision Recall
XGBoost algorithm 0.8689 0.6854 0.8436
Optimised XGBoost algorithm 0.9213 0.9412 0.8712
Test and analysis of model
Evaluation indicators

After administering the model, it is necessary to evaluate its performance. For the problem of binary classification, the model evaluation based on the confusion matrix is selected in this paper, and the evaluation indexes of the model are calculated, including the values of Accuracy, Precision and Recall of the model prediction, where the higher the index values of these three items, the better the prediction ability of the model, the higher the prediction accuracy. The calculation formula is as follows: Accuracy=TP+TNTP+TN+FP+FN {\rm{Accuracy}} = {{TP + TN} \over {TP + TN + FP + FN}} Precision=TPTP+FP {\rm{Precision}} = {{TP} \over {TP + FP}} Recall=TPTP+FN {\rm{Recall}} = {{TP} \over {TP + FN}} where TP is the number of positive values predicted correctly as positive values; FN is the number of positive values that are wrongly predicted as negative values; FP is the number of negative values that are wrongly predicted as positive values; and TN is the number of negative values correctly predicted as negative values. F=2Recall*PrecisionRecall+Precision F = {{2{\rm{Recall}}*{\rm{Precision}}} \over {{\rm{Recall}} + {\rm{Precision}}}} F value is defined as the harmonic average of Precision and’ Recall rate. When F value is high, it means that the test method is effective and the results are in line with the expectations.

Test results

Based on the evaluation indexes of the above models, the e-commerce customer loyalty prediction models based on XGBoost algorithm and improved XGBoost algorithm are compared, and the results are shown in Table 1.

The results shows that the optimised XGBoost model relies on strong feature learning ability, and can better extract user attributes and the information of commodity feature, so as to comprehensively capture the user's’ purchasing interest and get the user's’ shopping loyalty, whose Accuracy, Precision and Recall rates are 0.9213, 0.9412 and 0.8712, respectively. After calculation, the value of F of the e-commerce customer loyalty prediction model proposed in this paper is 0.9048, which is an excellent prediction result. It shows that aiming at the characteristics of missing data and unbalanced data in the experimental data set, the model can effectively avoid these problems and make the prediction result more accurate.

Conclusion

Based on the data mining technology, this paper constructs the prediction model of e-commerce customer loyalty by extracting the purchase records provided by a merchant in the Tmall platform, where the local abnormal factor algorithm is used to eliminate the data for cleaning, and XGBoost algorithm is used for optimisation. The results show that the prediction accuracy for e-commerce customer loyalty in the XGBoost model with improved penalty coefficient is higher, whose value of F is 0.9048. Therefore, the model can effectively solve the problems of missing experimental data sets and unbalanced data, so as to effectively extract the information of user attributes and commodity characteristics, and fully capture the purchase interest of users.

Fig. 1

Factors of e-commerce customer churn.
Factors of e-commerce customer churn.

Fig. 2

Classification of e-commerce customer loyalty.
Classification of e-commerce customer loyalty.

Fig. 3

Tasks of data mining technology in prediction of e-commerce customer loyalty.
Tasks of data mining technology in prediction of e-commerce customer loyalty.

Fig. 4

Data mining process of e-commerce platform.
Data mining process of e-commerce platform.

Fig. 5

Processing steps of e-commerce customer data.
Processing steps of e-commerce customer data.

Evaluation and comparison of e-commerce customer loyalty prediction models.

Accuracy Precision Recall
XGBoost algorithm 0.8689 0.6854 0.8436
Optimised XGBoost algorithm 0.9213 0.9412 0.8712

Tang Xin. Research on e-commerce user loyalty evaluation method [J]. Computer Simulation, 2017, 34(01):348–351:4–8. (in Chinese) XinTang Research on e-commerce user loyalty evaluation method [J] Computer Simulation 2017 34 01 348 351 4–8. (in Chinese) Search in Google Scholar

Wu Y, Li EY. Marketing mix, customer value, and customer loyalty in social commerce: a stimulus-organism-response perspective [J]. Internet Res, 2018, 28:7–10. WuY LiEY Marketing mix, customer value, and customer loyalty in social commerce: a stimulus-organism-response perspective [J] Internet Res 2018 28 7 10 10.1108/IntR-08-2016-0250 Search in Google Scholar

Wu Xiangyu. Application of data mining in the research of e-commerce customer behavior loyalty prediction [D]. Lanzhou University of Finance and Economics, 2016:7–23. (in Chinese) WuXiangyu Application of data mining in the research of e-commerce customer behavior loyalty prediction [D] Lanzhou University of Finance and Economics 2016 7 23 (in Chinese) Search in Google Scholar

Kim MJ, Hall CM. Can sustainable restaurant practices enhance customer loyalty? The roles of value theory and environmental concerns [J]. Journal of Hospitality and Tourism Management, 2020, 43:26–38. KimMJ HallCM Can sustainable restaurant practices enhance customer loyalty? The roles of value theory and environmental concerns [J] Journal of Hospitality and Tourism Management 2020 43 26 38 10.1016/j.jhtm.2020.03.004 Search in Google Scholar

Zhang Xiaonuo. Analysis of customer loyalty in e-commerce by using big data technology [J]. China Science and Technology Information, 2015(Z3):5–36. (in Chinese) ZhangXiaonuo Analysis of customer loyalty in e-commerce by using big data technology [J] China Science and Technology Information 2015 Z3 5 36 (in Chinese) Search in Google Scholar

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

Zhi Fenhe, Tian Ling, Ao Jinghai. Research on customer loyalty in e-commerce environment [J]. Business Research, 2007(1). (in Chinese) ZhiFenhe TianLing AoJinghai Research on customer loyalty in e-commerce environment [J] Business Research 2007 1 (in Chinese) Search in Google Scholar

Sun Hao, Ren Junsheng, Song Pingping. Marketing communication style and customer loyalty: a research summary. Modern Management Science, 2011, (1), 27–29. (in Chinese) SunHao RenJunsheng SongPingping Marketing communication style and customer loyalty: a research summary Modern Management Science 2011 1 27 29 (in Chinese) Search in Google Scholar

Ma Qian. Prediction of repeat customers of e-commerce platform based on machine learning [D]. Lanzhou University, 2017:10–35. (in Chinese) MaQian Prediction of repeat customers of e-commerce platform based on machine learning [D] Lanzhou University 2017 10 35 (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]. 2014, 10:476–494. ChenKuanchin HuYa-Han HsiehYi-Cheng Predicting customer churn from valuable B2B customers in the logistics industry: a case study [J]. 2014 10 476 494 10.1007/s10257-014-0264-1 Search in Google Scholar

Jones MA, Mothers Baugh DL, Beatty SE. Why customers stay: measuring the underlying dimensions of services switching costs and managing their differential strategic outcomes [J]. Journal of Business Research, 2002, 55(6):9–12. JonesMA Mothers BaughDL BeattySE Why customers stay: measuring the underlying dimensions of services switching costs and managing their differential strategic outcomes [J] Journal of Business Research 2002 55 6 9 12 10.1016/S0148-2963(00)00168-5 Search in Google Scholar

Li Min, Zhang Yuying, Yu Xinyu. Research on e-commerce customer loyalty based on cluster analysis [J]. Journal of Harbin University of Commerce (Natural Science Edition), 2017, 33(02):15–27. (in Chinese) LiMin ZhangYuying YuXinyu Research on e-commerce customer loyalty based on cluster analysis [J] Journal of Harbin University of Commerce (Natural Science Edition) 2017 33 02 15 27 (in Chinese) Search in Google Scholar

Tang Hu Xin. Research on e-commerce user loyalty evaluation method [J]. Computer Simulation, 2017, 34(01):3–8. (in Chinese) Hu XinTang Research on e-commerce user loyalty evaluation method [J] Computer Simulation 2017 34 01 3 8 (in Chinese) Search in Google Scholar

Ouyang Sheng. Application of data mining in consumer behavior analysis [D]. Hunan University, 2011. (in Chinese) ShengOuyang Application of data mining in consumer behavior analysis [D]. Hunan University 2011 (in Chinese) Search in Google Scholar

Wu Y, Li EY. Marketing mix, customer value, and customer loyalty in social commerce: a stimulus-organism-response perspective [J]. Internet Res, 2018, 28:7–10. WuY LiEY Marketing mix, customer value, and customer loyalty in social commerce: a stimulus-organism-response perspective [J] Internet Res 2018 28 7 10 10.1108/IntR-08-2016-0250 Search in Google Scholar

Dimyati M. The role of customer satisfaction in mediating marketing communication effect on customer loyalty [J]. Researchers World, 2015, 6:5–9. DimyatiM The role of customer satisfaction in mediating marketing communication effect on customer loyalty [J] Researchers World 2015 6 5 9 Search in Google Scholar

Oliveira T, Alhinho M, Rita P, et al. Modelling and testing consumer trust dimensions in e-commerce [J]. Computers in Human Behavior, 2017, 71:5–8. OliveiraT AlhinhoM RitaP Modelling and testing consumer trust dimensions in e-commerce [J] Computers in Human Behavior 2017 71 5 8 10.1016/j.chb.2017.01.050 Search in Google Scholar

Zhang Xiaonuo. Analysis of customer loyalty in e-commerce by using big data technology [J]. China Science and Technology Information, 2015(Z3):5–36. (in Chinese) ZhangXiaonuo Analysis of customer loyalty in e-commerce by using big data technology [J] China Science and Technology Information 2015 Z3 5 36 (in Chinese) Search in Google Scholar

Polecane artykuły z Trend MD

Zaplanuj zdalną konferencję ze Sciendo