1. bookAHEAD OF PRINT
Dettagli della rivista
License
Formato
Rivista
eISSN
2444-8656
Prima pubblicazione
01 Jan 2016
Frequenza di pubblicazione
2 volte all'anno
Lingue
Inglese
Accesso libero

Research on classification of e-commerce customers based on BP neural network

Pubblicato online: 01 Dec 2022
Volume & Edizione: AHEAD OF PRINT
Pagine: -
Ricevuto: 29 Mar 2022
Accettato: 14 Jun 2022
Dettagli della rivista
License
Formato
Rivista
eISSN
2444-8656
Prima pubblicazione
01 Jan 2016
Frequenza di pubblicazione
2 volte all'anno
Lingue
Inglese
Introduction

The economy in the e-commerce environment, based on informatisation and digitalisation, brings sellers and buyers into a brand-new information world without distance. Customers can obtain product-related information through various channels, ranging from passive consumption to fully rational analysis of product-related information, and then choose products or services that are really suitable for them. It can be said that customers in the e-commerce environment are becoming more and more mature [1]. On the other hand, with the transformation of the global economy, many industries are experiencing the problem of oversupply, forcing enterprises to reform their foothold from products to customers, which makes most businesses to involve in the fierce competition for customers [2]. In the e-commerce environment, the relationship between enterprises and customers has completely changed. In the fierce competition within the industry, the core management of the enterprise has gradually transformed from ‘product-centred’ to ‘customer-centred’, that is, only by meeting the customers’ need in a fastest and best way, can they continuously attract new customers and sustain old customers [3]. The first thing in the customer-centred approach is to classify customers reasonably. In the face of a large number of customer information and data, which are limited to superficial records and lack of in-depth analysis, it is indispensable to select a method for effective classification of customers [4].

The significance of this research is mainly reflected in three aspects: Firstly, applying the BP neural network to the actual classification of e-commerce customers can not only expand the application field of the BP neural network itself but also better analyse complicated factors in practical problems, which reflects the value of the BP neural network in the commercial field [5]. Secondly, the information of marketing activities is counted to predict the purchase tendency of consumers, so as to achieve accurate prediction of e-commerce customers, which results in the accurate prediction on the premise of extracting reasonable indicators of classification information. Finally, the improved customer classification model is used to distinguish customers with different purchasing tendencies for e-commerce, so that e-commerce companies can provide targeted, differentiated and effective marketing services for different customers, which not only reduces the cost but also improves the marketing profit and brings convenience to the management of relationship with e-commerce customers. Based on this, in this work, the main BP neural network algorithm is used to study the classification of e-commerce customers.

Overview of related theories
Theoretical basis of customer classification

The customers’ demands are infinite, and the limited resources owned by enterprises cannot meet the needs of all customers. In order to maximise social benefits as much as possible, enterprises must make rational use of limited resources and strive for high-value customers, which requires enterprises to have the ability to judge the value of customers. By implementing market segmentation by means of technology, enterprises can identify customers who can promote the development and then focus the limited resources on these potential customers [6]. According to the existing research, it is found that the measurement indexes of customer classification are diverse and the process is complicated, which mainly includes the following points: Firstly, enterprises should build an index system of customer classification. Different products and services of enterprises determine the difference in selected indicators. Therefore, enterprises should choose appropriate indicators according to the characteristics of their target customers [7]. At present, customer consumption involves a lot of information, including not only the basic attribute information of customers, such as customer gender, age and education level, but also other information of customer consumption, such as consumption frequency, total consumption amount and consumption satisfaction [8]. Secondly, enterprises should use big data to sort out the collected data, or other information technology methods to process the data; understand the customer consumption characteristics in a certain way; and calculate the customer value quantitatively [9]. Thirdly, according to the calculated customer value, enterprises should classify and manage customers and put forward appropriate suggestions for each type of customers to obtain maximum benefits [10]. The process of customer classification is shown in Figure 1.

Fig. 1

Specific process of customer classification.

Characteristics of customer classification in e-commerce environment

Compared with the characteristics of customer classification in the traditional business environment, customer classification in the e-commerce environment has changed a lot:

Classification indicators: Classification indicators are diversified. In the environment of e-commerce, enterprises classify customers not only by relying on customer value or some basic characteristics of customers but also by adding some log files left by customers on the Internet, from which personalised factors such as customers’ use feelings, psychological changes, preferences and habits can be analysed. These indicators can more realistically reflect the real needs of customers [11].

Customer-centred approach: In the environment of e-commerce, customers are becoming more and more mature and elusive and have transformed from passive consumption to active consumption, which is the main driver of transaction. Therefore, in order to attract new customers and retain old customers, enterprises must consider customers as the centre and classify customers. The classification results obtained in this way can help them design a complete set of network marketing schemes according to different types of customers [12].

Complete customer information: In the environment of e-commerce, the means for enterprises to obtain customer information has suddenly expanded with the application of the Internet. Enterprises can obtain detailed information and data of customers through various channels, which can truly reflect the real needs of customers [13].

Combining qualitative and quantitative analyses: In the e-commerce environment, in order to meet the increasing personalised, diversified and mature needs of customers, enterprises generally adopt the customer classification method combining qualitative and quantitative analyses to classify customers. This not only integrates the advantages of the two methods but also overcomes the shortcomings of the two methods, which make the classification result more reasonable [14].

Real-time information: In the e-commerce environment, due to the use of the Internet, the barriers of time and space have been broken. Enterprises and customers can use one-to-one, one-to-many and many-to-many online communications in real time, including text, voice and video, which results in more targeted information [15].

Establishment of index in e-commerce customer classification system
Basis of indicator selection

Choosing appropriate indicators is the key to the quality of customer classification results. A good classification indicator system should follow the following principles.

The scientific principle: The selection of indicators must have a theoretical basis, and indicators without practical significance cannot be selected. Generally speaking, the selection of indicators should follow the combination of science and practice, dynamic and static, and qualitative and quantitative [16].

The principle of comprehensiveness and independence: When selecting indicators, enterprises should try their best to consider them comprehensively from different aspects so as to keep the independence and representativeness of the indicators.

The principle of appropriateness: One of the principles that a qualified classification index must possess is appropriateness. When selecting customer classification indicators, it is necessary to have a clear distinction and reflect the specific information of customers. The appropriateness of the index must be judged in advance, and the corresponding verification should also be carried out for the customer classification. The principle of appropriateness not only reflects in the selection of indicators but also requires appropriateness in the employment of indicators.

The principle of measurability: A qualified classification index must be measurable to help study and compare the behaviours and attitudes of each customer group towards the specified products. When selecting indicators, it is better to avoid the indicator system, which is not easy to analyse or measure [17].

The principle of operability: The measurement and data collection of the customer classification index values should be feasible.

The principle of availability: The results of customer classification should be practical.

Generally, selecting the customer classification index in the e-commerce environment is an extremely critical and complicated process. In order to ensure the effective development of classification, it is necessary to combine the particularity and interactivity of enterprises to construct a classification indicator suitable for the actual situation of enterprises [1822].

Determination of classification indicators

This work will build a customer classification index system from the following three dimensions: characteristics of e-commerce customers, current values and potential values of the customers. The index system established from the aforementioned dimensions can also calculate the value of e-commerce customers more comprehensively and scientifically, as shown in Figure 2.

Fig. 2

Customer classification model based on customer value.

Characteristics of e-commerce customer

According to the existing research and practical basis, it can be found that the characteristics of e-commerce customers mainly include six secondary indicators: age, income, education level, geographical location, trust and customer interaction value. At the age of 26–40 years, the scale of monthly active users of Internet is the largest, and according to statistical data, it is also found that the middle-aged group is the main service target of e-commerce [23]. From the perspective of education level, the proportion of Internet users with junior high school education is the largest, followed by senior high school/technical secondary school education. As e-commerce is a new thing, the higher the education level, the greater the acceptance, so it can be considered that the customer value of those with a high education level is greater. From the perspective of income, Internet users with a monthly personal income of 3,001–5,000 yuan account for the largest proportion in China. Secondly, for the network users with 5,001–8,000 yuan and above 8,000 yuan, the consumption of e-commerce products especially needs the support of income, so it can be considered that the customer value above 8,000 yuan is relatively the largest [24]. Enhancing customers’ trust in the e-commerce platform and regional advantages and bringing good shopping experience to customers are conducive to enhance customers’ aspiration to consume. Therefore, it can be considered that customers with better geographical location have greater value.

Current value of customer

According to the existing research on the current value of customers, most scholars measure the benefits that customers bring to enterprises by applying the recency, frequency and monetary of them. Therefore, through the modification of RFM, the index for calculating the current value of customers is obtained. There are three specific indexes: firstly, the last time a customer spent on the e-commerce platform from now on, – generally, the shorter the time, the more enthusiastic the customer is about e-commerce consumption and the greater the customer value; secondly, the number of times that customers purchased on the e-commerce platform in the last year – the more times, the higher the dependence of customers on cross-border consumption and the greater the customer value; third, the total amount the customers spent on the e-commerce platform in the last year and the higher the amount the customers are willing to spend on the e-commerce platform – the more customers trust cross-border consumption, the greater the customer value [25].

Potential value of customers

The factors involved in the potential value of customers are quite complicated, including the structural changes of customers’ demands for products, which shows that customers not only repeatedly need purchased products but also may be interested in other products of enterprises. This also include customers’ psychological satisfaction with products and loyalty to enterprises, which mainly show that customers prefer this kind of products, customers love the product design, and culture publicity is effective [26]. Generally speaking, the more the customers’ demands for enterprise products and the more satisfied and loyal they are to enterprise products, the greater the customer value [27]. The customer potential value built in this study includes seven secondary indicators, namely, the number of repeated purchases for customers, the number of cross-purchases by customers, the durability of the relationship between customers and enterprises, the degree of customers’ perception of enterprise brand image, whether they are willing to recommend others to buy, customers’ scores and customers’ willingness to become members.

Establishment of index system

The e-commerce customer classification index system is constructed from the customer characteristics, current value and potential value, which comprehensively considers all factors that may affect the customer value, and it is scientific to calculate the customer value with this index system. According to the existing literature research, realistic statistical data and cases, an e-commerce customer classification index system is established, which is shown in Table 1.

Classification index system about e-commerce customer based on customer value.

Primary indexSecondary index
Customer age
Customer income level
Customer characteristics of e-commerceCustomer education
Geographical position
Degree of trust
Customer interaction value
Time from the customers’ last e-commerce consumption to now
Current valueNumber of e-commerce consumption of customers in the last year
Total amount spent by customers on e-commerce consumption in the last year
Number of customer cross purchases
Persistence of customer enterprise relationship
Potential valueCustomers’ perception of corporate brand image
Are you willing to recommend others to buy? Customer score
Customers’ willingness to become members
Number of customer cross-purchases
Calculation of index weight

Weight setting

The three indicators in the current customer value reflect the different importance of customers to the enterprise. Accurate determination of the weight of each specific indicator plays an important role in customer pre-classification in the current value. At present, the methods to determine the index weight include expert grading method and analytic hierarchy process. In this study, the expert grading method is used to establish the importance of the index, and the corresponding weight set is a={a1,a2,,an} , and Σai=1 . After the index weight is determined, the customer’s current value can be calculated according to the corresponding index data of each sample, as shown in Figure 3.

Calculate the current value score of each customer.

Through the five-point scoring method, the corresponding numerical values of the three indicators in the current value are known and these numerical values are multiplied by the weight of each indicator that is established by the corresponding expert grading method, respectively. Finally, the current value of the customer can be calculated by adding up, as given the following formula: Ci=Ri×WR+Fi×WF+Mi×WM where Ci represents the current value of each sample customer; Ri, Fi and Mi, respectively, represent the scores of the sample customer in the three indicators; and WR, WF and WM, respectively, represent the weight of the sample customer in the three indicators.

Pre-classification of e-commerce customers

In this study, the five-point scoring method has been used to establish the scores that customers can obtain for each index. For the sake of simplicity of calculation, customers are pre-divided into five categories, and the higher the score, the greater the current value of customers.

Fig. 3

Customer’s current value weight.

Design of e-commerce customer classification model based on BP neural network
BP neural network algorithm

The BP neural network is a multi-layer feedforward neural network, which consists of input layer, hidden layer and output layer [28]. The learning process of the network consists of forward propagation and backward propagation. In the forward propagation, the data pass through the input layer, the hidden layer and the output layer in turn. If the result fails to meet the expected requirements, it will enter the backward propagation, where it distributes the error to each small unit in the network structure, and through constant adjustment, until the output result reaches the expectation [29].

When calculating each node, it needs to use the S-type function for basic calculation, so it requires a hidden layer to solve the problem of decision classification. When there are two hidden layers, the output function can be used on the input image at will. However, for a small network, one hidden layer is enough to complete the work [30].

The actual output is calculated in the direction from input to output, while the weights and thresholds are corrected in the direction from output to input [31].

xj represents the input of the jth node of the input layer, j=1,.,m ;

wij represents the weight between the ith node of the hidden layer and the jth node of the input layer;

θi represents the threshold of the ith node of the hidden layer;

ϕ(x) represents the excitation function of the hidden layer;

wki represents the weight between the kth node of the output layer and the ith node of the hidden layer, with i=1,.,q ;

ak represents the threshold of the kth node of the output layer, k=1,.,l ;

Ψ(x) represents the excitation function of the output layer; and

Ok represents the output of the kth node of the output layer.

(1) Forward propagation of the signal

Input neti of the ith node of the hidden layer: neti=j=1Mwijxj+θi

Output y of the ith node of the hidden layeri: yi=ϕ(neti)=ϕ(j=1Mwijxj+θi) $${y_i} = \phi (ne{t_i}) = \phi \left( {\mathop \sum \limits_{j = 1}^M {w_{ij}}{x_j} + {\theta _i}} \right)$$

Output the input net of the kth node of the layerk: netk=i=1qwkiyi+ak=i=1qwkiϕ(j=1Mwijxj+θi)+ak $$ne{t_k} = \mathop \sum \limits_{i = 1}^q {w_{ki}}{y_i} + {a_k} = \mathop \sum \limits_{i = 1}^q {w_{ki}}\phi \left( {\mathop \sum \limits_{j = 1}^M {w_{ij}}{x_j} + {\theta _i}} \right) + {a_k}$$

Output o of the kth node of the output layer: ok=ψ(netk)=ψ(i=1qwkiyi+ak)=ψ(i=1qwkiϕ(j=1Mwijxj+θi)+ak) $${o_k} = \psi (ne{t_k}) = \psi \left( {\mathop \sum \limits_{i = 1}^q {w_{ki}}{y_i} + {a_k}} \right) = \psi \left( {\mathop \sum \limits_{i = 1}^q {w_{ki}}\phi \left( {\mathop \sum \limits_{j = 1}^M {w_{ij}}{x_j} + {\theta _i}} \right) + {a_k}} \right)$$

(2) Backward propagation of error

Backward propagation of error, that is, firstly, the output error of neurons in each layer is calculated layer by layer from the output layer, and then the weights and thresholds of each layer are adjusted according to the error gradient descent method, so the final output of the modified network can approach the expected value.

The quadratic error criterion function for each sample P is Ep: Ep=12k=1L(Tkok)2

The total error criterion function of the system for P training samples is E=12p=1Pk=1L(Tkpokp)2

According to the error gradient descent method, the output layer weight correction Δwki, output layer threshold correction Δak, hidden layer weight correction Δwij and hidden layer threshold correction Δθi are considered as follows: Δwki=ηEwki;Δak=ηEak;Δwij=ηEwij;Δθi=ηEθi $${\rm{\Delta }}{w_{ki}} = - \eta {{\partial E} \over {\partial {w_{ki}}}};\>{\rm{\Delta }}{a_k} = - \eta {{\partial E} \over {\partial {a_k}}};\>{\rm{\Delta }}{w_{ij}} = - \eta {{\partial E} \over {\partial {w_{ij}}}};\>{\rm{\Delta }}{\theta _i} = - \eta {{\partial E} \over {\partial {\theta _i}}}$$

The weight adjustment formula of the output layer: Δwki=ηEwki=ηEnetknetkwki=ηEokoknetknetkwki

The output layer threshold adjustment formula: Δak=ηEak=ηEnetknetkak=ηEokoknetknetkak

The weight adjustment formula of the hidden layer: Δwij=ηEwij=ηEnetinetiwij=ηEyiyinetinetiwij

The hidden layer threshold adjustment formula: Δθi=ηEθi=ηEnetinetiθi=ηEyiyinetinetiθi Eok=p=1Pk=1L(Tkpokp) netkwki=yi,netkak=1,netiwij=xj,netiθi=1 $$\matrix{ {{{\partial ne{t_k}} \over {\partial {w_{ki}}}}} & { = {y_i},\>{{\partial ne{t_k}} \over {\partial {a_k}}} = 1,\>{{\partial ne{t_i}} \over {\partial {w_{ij}}}} = {x_j},{{\partial ne{t_i}} \over {\partial {\theta _i}}} = 1} \cr } $$ Eyi=p=1Pk=1L(Tkpokp)ψ(netk)wki yineti=ϕ(neti) oknetk=ψ(netk)

The following formula is finally obtained: Δwki=ηp=1Pk=1L(Tkpokp)ψ(netk)yi Δak=ηp=1Pk=1L(Tkpokp)ψ(netk) Δwij=ηp=1Pk=1L(Tkpokp)ψ(netk)wkiϕ(neti)xj Δθi=ηp=1Pk=1L(Tkpokp)ψ(netk)wkiϕ(neti)

The BP network is a one-way multi-layer feedforward network, which includes input layer, hidden layer and output layer. It is a widely used model at present. The algorithm adopts error backward propagation learning method in the hierarchical network structure, and the learning process consists of forward propagation and error back propagation.

The algorithm of the BP neural network is a kind of learning algorithm that has the characteristics of supervision. The main idea is that for q input learning samples P1, P2,., PQ, the expected output samples corresponding to them are known as T1, T2,., TQ. The learning algorithm trains the algorithm and corrects the weights by comparing the actual outputs A1, A2, ⋯⋯ of the network with the gaps between the targets T1, T2,., TQ, which makes A and T as close as possible (Figure 4).

Fig. 4

BP neural network model.

Establishment of classification model based on BP neural network

The three-layer network structure is often used in the research of customer classification in the BP neural network algorithm, so three-layer BP neural network structure is selected to calculate customer value, and the preparatory work includes determining the number of network layers, the number of input layer nodes, the number of output layer nodes, the number of hidden layer nodes and data processing.

Determine the number of layers of BP neural network

In this work, it is found that the research on customer classification only involves the index data at the input level and the measurement of customer value at the output level, and only the BP neural network model with one hidden level can measure customer value, so this study selected the three-layer BP neural network model to measure customer value.

Determine the number of neurons in the input layer

The customer classification index system constructed in this work involves a total of 16 specific indicators, and the benefits that customers can bring to enterprises are calculated according to these indicators, so the number of neurons in the input layer of customer value calculation is 16.

Determine the number of neurons in the hidden layer

The determination of hidden layers is very significant in BP neural network training, which will affect the final calculation result, but the number of hidden layers is not constant. Generally, the more neurons in the input layer or the output layer, the more complex the network structure will be, and the more neurons in the hidden layer will be needed. Specifically, the number of neurons in the hidden layer to be selected can be determined according to the training results in actual operation.

The relationship between the number of neurons in the hidden layer and the number of neurons in the input layer and output layer in the common three-layer BP neural network model is as follows: j=i+k+a where j is the number of neurons in the hidden layer, i is the number of neurons in the input layer, k is the number of neurons in the output layer and a is any natural number from 1 to 10.

Determine the number of neurons in the output layer

The basis of customer classification in this study is mainly the size of customer value, so in empirical research, the BP neural network is used to calculate customer value, and there is only one output value involved in each sample, that is, customer value, which is expressed by K.

Data processing

When training the BP neural network, it is necessary to select certain sample data in advance for training, simulation test and verification, and to normalise the data that have been scored with a five-point system to minimise and maximise, so that the data used for network training are within the range of [0, 1]. The equation used is as follows: Xi=XiMIN(X)MAX(X)MIN(X) where Xi represents a single individual in a sample; MIN(X) and MAX(X) represent the minimum and maximum values in a single individual of the sample, respectively; and Xi represents the single individual data of the sample after normalisation.

Then, the classification model of the e-commerce customer based on the BP neural network is constructed as follows (Figure 5):

Fig. 5

BP neural network model diagram of customer value evaluation.

Analysis of results

After the customer pre-classification processing is completed, half of the customers in each category are randomly sampled as samples for customer classification research, which constitutes a set of data with 260 samples. The reason for further processing is that the samples trained by the BP neural network cannot be too much. Selecting customers from each pre-classification category as sample capacity for training is beneficial to improving the representativeness and balance of samples. Then, the sampled sample data are normalised by minimum and maximum, so that their sizes are all within the range of [0, 1].

Establishment of neural network

The sample data are divided into three categories for training, simulation test and verification of the neural network, while the specific function of the sample is randomly determined by the system. Here, 70% of customer data are taken as the training sample, and the sample size is 182; 15% of customer data are taken as the verification sample, and the sample size is 39; 15% customer data are taken as the test sample, and the sample size is 39. In addition, the number of neurons in the hidden layer is determined; here, it is the default value of 10. Generally, after the default value is selected, the number of neurons can be adjusted according to the actual situation, and different training results can be evaluated, and the number of hidden layers with the smallest error can be selected [32]. Moreover, the number of hidden layers is 6, 7, 8, 9, 10, 11 and 12, respectively. After comparing the mean square error, it is determined that the number of hidden layers is 10. At the same time, according to the network structure diagram, the number of neurons in the input layer is 16 and the number of neurons in the output layer is 1.

Training, verification and testing

Before the neural network training, the maximum training times, the maximum error value and the gradient of the error surface will be set in advance. When one or more indexes reach the set value during the network training, the training will end. Training times is set to the maximum number of iterations, that is, 1000 times, and the final result shows 105 iterations. During the training, the result shows that the training ends in 0.06 seconds, with an initial error of 0.183; at the same time, the initial gradient error of the error surface is 0.368; the maximum number of verification times is 6, and if it exceeds that number, the training will stop, which represents that the training error is greater than the expected error, and the validity verification will stop at once (Table 2).

Training results.

Potential value training times1,000
Training time0.06 s
Error0.183
Gradient of error surface0.368
Verification times6
Fitting evaluation

According to the results, the smaller the value of MSE (performance function), the closer the R value is to 1 and the better the training effect. The results are as follows (Table 3).

Fitting evaluation results.

SampleMSER
Train1821.15322e-129.99999e-1
Verification392.40519e-69.99985e-1
Test391.80075e-59.99899e-1

Figure 6 shows the error convergence of network training, in which the abscissa represents the training times and the ordinate represents the error value. This training converges at the 105th time, stops training and reaches the minimum error value at the 104th time, which converges to 2.4052 × 10–6.

Fig. 6

Mean square error.

Result analysis

According to the empirical analysis, it can be found that the calculation accuracy of customer value is high. In this study, the trained network has been named net and saved in the workspace of MATLAB software. In practical application, the customer value can be calculated only by using the Subscriber Identity Module (SIM) function [33]. The function of SIM is Y = SIM (X, net), where X represents the numerical value reflected by each index and Y is the calculated customer value. The trained neural network model is used to calculate the customer value that is sorted according to the value, which is divided into four categories, so as to facilitate the subsequent enterprises to formulate targeted customer management measures (Table 4).

Customer value classification results.

Customer categoryCustomer value
Class A customer[0.75, 1]
Class B customer[0.5, 0.75]
Class C customer[0.25, 0.5]
Class D customer[0, 0.25]

The concrete result is that the value of customers in class A is the largest, and the value range is within the range of [0.75, 1]. This kind of customers create the greatest benefits for enterprises, and they will actively purchase from the e-commerce platform and may become loyal customers without excessive publicity and promotion. Customers in class B are of great value, and the value range is within the range of [0.5, 0.75]. This kind of customers can also bring great benefits to enterprises, are interested in e-commerce products and have a higher probability of transforming their value into customers in class A, The value of customers in class C is small, and the numerical range is within the range of [0.25, 0.5]. These customers are sceptical about the way of e-commerce consumption, and they are still in the initial stage of understanding the products of the platform, so the input-to-output ratio of enterprises on this customer will not be very large. Customers in class D have the smallest value, and the value range is within the range of [0, 0.25]. This kind of customers do not trust the consumption pattern of e-commerce, and the benefits they can create for enterprises are very limited. Enterprises are advised to give up on them because they have high requirements on products but show little consumption. Even if enterprises invest high resources in customer development, the customer value they can obtain is still very small.

Conclusion

In this study, the BP neural network algorithm is used to classify e-commerce customers. First of all, the customer data of Tmall were obtained by a questionnaire survey, and the data were sorted out and normalised. Secondly, in order to improve the representativeness of the sample, customers were pre-divided into five categories based on the current value data of customers, and half of the customer data from each category were randomly sampled as the sample size of empirical research. Finally, the input and output data of the sample were trained by the BP neural network to calculate the customer value, which was divided into four categories according to the value.

Fig. 1

Specific process of customer classification.
Specific process of customer classification.

Fig. 2

Customer classification model based on customer value.
Customer classification model based on customer value.

Fig. 3

Customer’s current value weight.
Customer’s current value weight.

Fig. 4

BP neural network model.
BP neural network model.

Fig. 5

BP neural network model diagram of customer value evaluation.
BP neural network model diagram of customer value evaluation.

Fig. 6

Mean square error.
Mean square error.

Customer value classification results.

Customer category Customer value
Class A customer [0.75, 1]
Class B customer [0.5, 0.75]
Class C customer [0.25, 0.5]
Class D customer [0, 0.25]

Fitting evaluation results.

Sample MSE R
Train 182 1.15322e-12 9.99999e-1
Verification 39 2.40519e-6 9.99985e-1
Test 39 1.80075e-5 9.99899e-1

Classification index system about e-commerce customer based on customer value.

Primary index Secondary index
Customer age
Customer income level
Customer characteristics of e-commerce Customer education
Geographical position
Degree of trust
Customer interaction value
Time from the customers’ last e-commerce consumption to now
Current value Number of e-commerce consumption of customers in the last year
Total amount spent by customers on e-commerce consumption in the last year
Number of customer cross purchases
Persistence of customer enterprise relationship
Potential value Customers’ perception of corporate brand image
Are you willing to recommend others to buy? Customer score
Customers’ willingness to become members
Number of customer cross-purchases

Training results.

Potential value training times 1,000
Training time 0.06 s
Error 0.183
Gradient of error surface 0.368
Verification times 6

Ling J K, Chiu C C. Mining the customer credit by using the neural network model with classification and regression tree approach[C]//IFSA World Congress and 20th NAFIPS International Conference, 2001. Joint 9th. IEEE, 2001. Ling J K, Chiu C C. Mining the customer credit by using the neural network model with classification and regression tree approach[C] //IFSA World Congress and 20th NAFIPS International Conference, 2001. Joint 9th. IEEE, 2001.Search in Google Scholar

Yang G, Yuan X C. Bank customer classification model based on elman neural network optimized by PSO [J]. IEEE, 2007:5667–5670. Yang G, Yuan X C. Bank customer classification model based on elman neural network optimized by PSO [J]. IEEE, 2007:56675670.10.1109/WICOM.2007.1390Search in Google Scholar

Velu C M, Kashwan K R. Artificial neural network based data mining technique for customer classification for market forecasting. 2015. Velu C M, Kashwan K R. Artificial neural network based data mining technique for customer classification for market forecasting. 2015.Search in Google Scholar

Wan Y H, Wan-Ping H U, Cao X P. Constructing customer consumption classification models based on rough sets and neural network [J]. Journal of Industrial Engineering and Engineering Management, 2011. Wan Y H, Wan-Ping H U, Cao X P. Constructing customer consumption classification models based on rough sets and neural network [J]. Journal of Industrial Engineering and Engineering Management, 2011.Search in Google Scholar

Yong L. Research on customer classification basing on neural network [J]. Value Engineering, 2004. Yong L. Research on customer classification basing on neural network [J]. Value Engineering, 2004.Search in Google Scholar

Rani K L, Loshma G. Classification and prediction in CRM using back propagation multilayer feed forward neural network approach [J]. International Archive of Applied Sciences & Technology, 2012. Rani K L, Loshma G. Classification and prediction in CRM using back propagation multilayer feed forward neural network approach [J]. International Archive of Applied Sciences & Technology, 2012.Search in Google Scholar

H işik, Arslan S. An artificial neural network classification approach for use the ultrasound in physiotherapy [J]. Journal of Medical Systems, 2011, 35(6):1333–1341. H işik, Arslan S. An artificial neural network classification approach for use the ultrasound in physiotherapy [J]. Journal of Medical Systems, 2011, 35(6):13331341.10.1007/s10916-009-9410-620703780Search in Google Scholar

Jie Z, Jie L U, Zhang G, et al. An integrated analysis method for bank customer classification [M]. Jie Z, Jie L U, Zhang G, An integrated analysis method for bank customer classification [M].Search in Google Scholar

Ahn H, Ahn J J, Oh K J, et al. Facilitating cross-selling in a mobile telecom market to develop customer classification model based on hybrid data mining techniques [J]. Expert Systems with Applications, 2011, 38(5):5005–5012. Ahn H, Ahn J J, Oh K J, Facilitating cross-selling in a mobile telecom market to develop customer classification model based on hybrid data mining techniques [J]. Expert Systems with Applications, 2011, 38(5):50055012.10.1016/j.eswa.2010.09.150Search in Google Scholar

Cai Liyan, Du Yongyuan. Research on customer churn analysis in mobile communication industry based on data mining [J]. Microcomputer Information, 2010, 26(30):154–156. Cai Liyan, Du Yongyuan. Research on customer churn analysis in mobile communication industry based on data mining [J]. Microcomputer Information, 2010, 26(30):154156.Search in Google Scholar

Pitner, Tomáš, Kriksciuniene D, Sakalauskas V. Tracking customer portrait by unsupervised classification techniques [J]. Transformations in Business and Economics, 2012, 11(3):167–189. Pitner, Tomáš, Kriksciuniene D, Sakalauskas V. Tracking customer portrait by unsupervised classification techniques [J]. Transformations in Business and Economics, 2012, 11(3):167189.Search in Google Scholar

Yan Z, Wang X L, Li C Q. Study based on “Situational Rationality” hypothesis for customer market classification model [C]//The Third International Conference on Operations and Supply Chain Management. Inner Mongolia Finance and Economics College, Hohhot, P.R.China; School of Management, Inner Mongolia University of Technology Hohhot, P.R.China, 010051; 2009. Yan Z, Wang X L, Li C Q. Study based on “Situational Rationality” hypothesis for customer market classification model [C]//The Third International Conference on Operations and Supply Chain Management. Inner Mongolia Finance and Economics College, Hohhot, P.R.China; School of Management, Inner Mongolia University of Technology Hohhot, P.R.China, 010051; 2009.Search in Google Scholar

Anzhi, Qi. An unsupervised neural network algorithm SOMO for continuous optimization problems[C]//. Anzhi, Qi. An unsupervised neural network algorithm SOMO for continuous optimization problems[C]//.Search in Google Scholar

Zhong J, Li W. Predicting customer call intent by analyzing phone call transcripts based on CNN for multi-class classification [J]. 2019. Zhong J, Li W. Predicting customer call intent by analyzing phone call transcripts based on CNN for multi-class classification [J]. 2019.10.5121/csit.2019.90702Search in Google Scholar

Nazaraghaei M, Ghiasi H, Chafi M A. Classification of customer’s credit risk using ensemble learning (case study: sepah bank) (in Persian) [J]. Journal of Monetary and Banking Research, 2019, 12. Nazaraghaei M, Ghiasi H, Chafi M A. Classification of customer’s credit risk using ensemble learning (case study: sepah bank) (in Persian) [J]. Journal of Monetary and Banking Research, 2019, 12.Search in Google Scholar

Moudani W, Zaarour G, Mora-Camino F. Fuzzy classification of customer insolvency in mobile telecommunication [J]. International Journal of Decision Support System Technology, 2014. Moudani W, Zaarour G, Mora-Camino F. Fuzzy classification of customer insolvency in mobile telecommunication [J]. International Journal of Decision Support System Technology, 2014.10.4018/ijdsst.2014070101Search in Google Scholar

Abumallouh A, Qawaqneh Z, BD Barkana. Deep neural network combined posteriors for speakers’ age and gender classification[C]//Industrial Electronics, Technology & Automation. IEEE, 2017. Abumallouh A, Qawaqneh Z, Barkana BD. Deep neural network combined posteriors for speakers’ age and gender classification[C]//Industrial Electronics, Technology & Automation. IEEE, 2017.10.1109/CT-IETA.2016.7868251Search in Google Scholar

Hu W, Yu M. The study of customer classification based on self-organizing feature map neural network[C]//2010 International Conference on Artificial Intelligence and Education (ICAIE). IEEE, 2010. Hu W, Yu M. The study of customer classification based on self-organizing feature map neural network[C]//2010 International Conference on Artificial Intelligence and Education (ICAIE). IEEE, 2010.Search in Google Scholar

Cao Y. A Customer classification based modified fuzzy neural network [J]. Computer Engineering and Applications, 2006. Cao Y. A Customer classification based modified fuzzy neural network [J]. Computer Engineering and Applications, 2006.Search in Google Scholar

Zhou L, Wu Q Z. Application of GA-BP algorithm in customer classification. IEEE, 2011:255–257. Zhou L, Wu Q Z. Application of GA-BP algorithm in customer classification. IEEE, 2011:255257.10.1109/ICSESS.2011.5982302Search in Google Scholar

Hodecker A, Fernandes A, Steffens A, et al. Clothing classification using convolutional neural networks [C]//2020 15th Iberian Conference on Information Systems and Technologies (CISTI). 2020. Hodecker A, Fernandes A, Steffens A, Clothing classification using convolutional neural networks [C]//2020 15th Iberian Conference on Information Systems and Technologies (CISTI). 2020.10.23919/CISTI49556.2020.9141035Search in Google Scholar

Song W, Lei H. Research on port customer classification based on RBF neural network[C]//2015 International Conference on Logistics, Informatics and Service Sciences (LISS). IEEE, 2015. Song W, Lei H. Research on port customer classification based on RBF neural network[C]//2015 International Conference on Logistics, Informatics and Service Sciences (LISS). IEEE, 2015.Search in Google Scholar

Wang J, Wang C, Wang Z. The study of power customer classification based on principal component analysis and improved back propagation neural network [C]//International Conference on Biomedical Engineering & Informatics. IEEE, 2009. Wang J, Wang C, Wang Z. The study of power customer classification based on principal component analysis and improved back propagation neural network [C]//International Conference on Biomedical Engineering & Informatics. IEEE, 2009.10.1109/BMEI.2009.5304468Search in Google Scholar

Gheorghe G, Ivanov O, Gavrilas M. Customer classification and load profiling using data from smart meters [C]//Symposium on Neural Network Applications in Electrical Engineering. IEEE, 2015. Gheorghe G, Ivanov O, Gavrilas M. Customer classification and load profiling using data from smart meters [C]//Symposium on Neural Network Applications in Electrical Engineering. IEEE, 2015.Search in Google Scholar

Jihan N, Senarath Y, Ranathunga S. Aspect extraction from customer reviews using convolutional neural networks [C]//ICTer 2019. 2018. Jihan N, Senarath Y, Ranathunga S. Aspect extraction from customer reviews using convolutional neural networks [C]//ICTer 2019. 2018.10.1109/ICTER.2018.8615575Search in Google Scholar

Yosef I, Samuels C I. Multifactor customer classification model for IP transit product [C]//2014 8th International Conference on Telecommunication Systems Services and Applications (TSSA). IEEE, 2014. Yosef I, Samuels C I. Multifactor customer classification model for IP transit product [C]//2014 8th International Conference on Telecommunication Systems Services and Applications (TSSA). IEEE, 2014.10.1109/TSSA.2014.7065955Search in Google Scholar

Li X. A new customer classification algorithm for electronic commerce enterprises [J]. Journal of Digital Information Management, 2012, 10(5):284–288. Li X. A new customer classification algorithm for electronic commerce enterprises [J]. Journal of Digital Information Management, 2012, 10(5):284288.Search in Google Scholar

Hu W, Pei Z. After-sales service oriented automobile customer classification index system and evaluation example. IEEE, 2011. Hu W, Pei Z. After-sales service oriented automobile customer classification index system and evaluation example. IEEE, 2011.Search in Google Scholar

Ting, Wu, Hui, et al. A multi-factor customer classification evaluation model [J]. Computing and Informatics, 2010, 29(4):509–520. Ting, Wu, Hui, A multi-factor customer classification evaluation model [J]. Computing and Informatics, 2010, 29(4):509520.Search in Google Scholar

Chicco G, Napoli R, Piglione F, et al. Emergent electricity customer classification [J]. IEE Proceedings - Generation, Transmission and Distribution, 2005, 152(2):164–172. Chicco G, Napoli R, Piglione F, Emergent electricity customer classification [J]. IEE Proceedings - Generation, Transmission and Distribution, 2005, 152(2):164172.10.1049/ip-gtd:20041243Search in Google Scholar

Liu C H, Mei Q, Cai S Q. The model of customer classification and evaluation of customer value based on RFM [J]. Technoeconomics & Management Research, 2012. Liu C H, Mei Q, Cai S Q. The model of customer classification and evaluation of customer value based on RFM [J]. Technoeconomics & Management Research, 2012.Search in Google Scholar

Lopez J J, Aguado J A, Martin F, et al. Electric customer classification using nopfield recurrent ANN [C]//Electricity Market, 2008. EEM 2008. 5th International Conference on European. IEEE, 2008. Lopez J J, Aguado J A, Martin F, Electric customer classification using nopfield recurrent ANN [C]//Electricity Market, 2008. EEM 2008. 5th International Conference on European. IEEE, 2008.10.1109/EEM.2008.4579053Search in Google Scholar

Xiao J, Cao H, Jiang X, et al. GMDH-based semi-supervised feature selection for customer classification [J]. Knowledge-Based Systems, 2017, 132(sep.15):236–246. Xiao J, Cao H, Jiang X, GMDH-based semi-supervised feature selection for customer classification [J]. Knowledge-Based Systems, 2017, 132(sep. 15):236246.10.1016/j.knosys.2017.06.018Search in Google Scholar

Articoli consigliati da Trend MD

Pianifica la tua conferenza remota con Sciendo