The Construction of Consumer Buying Behavior Analysis System

Based on Data Mining in Social Marketing

Di Zhang

1,2

, Dongphil Chun

and Minghao Huang

Pukyong National University, Graduate school of Management of Technology, Busan, 48548, Korea

Economic and Management School, Hulunbuir University, Inner Mongolia Hulunbuir 021008, China

Seoul School of Integrated Sciences & Technologies, Seoul, 03767, Korea

Keywords: Data Mining, Python, Social Marketing, Analysis of Purchasing Behavior, Machine Learning Algorithm

Model.

Abstract: This paper takes data mining technology as the core, completes the design and construction of machine

learning models such as K-mediod algorithm, Decision Tree algorithm and Apriori algorithm by using

Numpy module in Python language environment, and realizes the development of user classification, feature

prediction and correlation analysis of consumer purchasing behavior data under social marketing. It com-

bines Django development framework to realize the integration and encapsulation of various functional

modules, and finally forms a web-based consumer purchasing behavior analysis system. The system adopts

B/S architecture, and completes the deployment of all levels of the system and the planning and design of

business logic according to MVC mode. It can facilitate users to discover the influence of social marketing,

product attributes, service methods and other factors on consumers' purchasing behavior through concise

and efficient operation, and take reasonable measures to improve social marketing strategies, adjust service

methods, enrich service content, and maximize the benefits of enterprises.

1 INTRODUCTION

With the continuous change and improvement of

network information technology, the social market-

ing is a "new marketing" model based on social

relations, which uses the daily social activities of

social media users and the dissemination of their

own content to spread the brand information of en-

terprises. The social marketing will no longer focus

on products, but turn to consumers, and its content

and form are obviously different from the traditional

marketing. More information outside the product,

such as application scenarios, content expression,

community groups, and psychological sense of be-

longing, will be relied on to promote the generation

of consumers' purchasing behavior. So, the logical

chain of consumer buying behavior has changed

from "attention, choice and purchase" in the tradi-

tional marketing mode to "sharing, purchasing and

paying attention". (Liu, 2021)

For enterprises, how to choose a suitable plat-

form, formulate a perfect social marketing strategy,

and realize the interaction, locking and transfor-

mation with the target audience has become the key

to "stand out". So, this paper thinks that taking data

mining technology as the core, in Python language

environment, using Django frame architecture to

complete the construction of consumer buying be-

havior analysis system. Through the analysis and

mining of consumers' purchasing behavior, the sys-

tem enables the e-commerce platform to compre-

hensively obtain consumers' real needs, develop

customer value, improve customer service quality,

and provide guidance for the formulation, imple-

mentation and management of its social marketing

strategy, and finally realize the growth and efficien-

cy of enterprises.

2 OVERVIEW OF KEY

TECHNOLOGIES

2.1 Data Mining Technology

As a kind of computer science and technology, data

mining technology aiming at the complex process of

extracting and mining hidden and valuable patterns

382

Zhang, D., Chun, D. and Huang, M.

The Construction of Consumer Buying Behavior Analysis System Based on Data Mining in Social Marketing.

DOI: 10.5220/0011737800003607

In Proceedings of the 1st International Conference on Public Management, Digital Economy and Internet Technology (ICPDI 2022), pages 382-385

ISBN: 978-989-758-620-0

 2023 by SCITEPRESS – Science and Technology Publications, Lda. Under CC license (CC BY-NC-ND 4.0)

or laws from a large number of incomplete and noisy

actual data. (Zhang, 2021) The realization of general

data mining involves three steps: data preparation,

data mining and result presentation. In the stage of

data mining, we will focus on the selection of data

samples, sample features and the establishment of

algorithm models.

2.2 Python

The Python is a high-level scripting language that

combines interpretive, compiler, interactive and

object-oriented. 3]. The Django is a custom Web

development framework based on Python. Its overall

structure is similar to MVC pattern, but not exactly

the same as MVC pattern. Under Django framework,

we focus on Model, Template and Views, also

known as MTV pattern. (Wang, 2021)

2.3 Development Process

In the data preparation stage, the data collection

comes from the data of consumers' purchasing be-

havior in the enterprise's own e-commerce platform

on the one hand, and the feedback from the social

questionnaire on the other hand. After

pre-processing, the two kinds of data will collect a

total of 56 attribute variable values from three as-

pects: consumers, e-commerce and social influence,

covering the whole process of e-commerce. In the

stage of data mining, according to the data of con-

sumers' purchasing behavior and questionnaire, three

attribute variables of purchasing frequency, pur-

chasing amount and latest purchasing time are se-

lected, and the RFM model of K-mediod algorithm

is adopted to divide consumers into three categories:

important value consumers, general value consumers

and worthless consumers. The key code of function

realization is shown in Figure 1.

In the face of numerous attribute data, based on

the calculation results of the dependency model, 10

key attributes are selected as the input attributes of

feature prediction, and the purchase behavior rules

of different types of consumers are further mined

through the Decision Tree algorithm model. The

implementation of Decision Tree algorithm model

will be realized by C4.5 algorithm, and the key code

is shown in Figure 2.

Figure 1: Cluster analysis results of K-Mediod algorithm [Original].

The Construction of Consumer Buying Behavior Analysis System Based on Data Mining in Social Marketing

383

Figure 2: Key code of Decision Tree model implemented by C4.5 algorithm [Original].

Figure 3: Consumer cluster analysis results of RFM model [Source: Network].

After further accurate acquisition of the consumer

purchasing behavior rules, the future purchasing

behavior will be predicted from the correlation rules

analysis of the products, services and the marketing

methods concerned about. The association rules are

realized by Apriori algorithm, in which the key

support and confidence indexes determine the

strength of the association, and are used to guide the

enterprise's social marketing decision and scheme

formulation.

The Web application server is implemented by

Django framework, the operating system is Win-

dows10.0. The Web server is Nginx server, the pro-

ject development language is Python 3.6.7, the de-

velopment tool is PyCharm 2018.3.1 x64, and the

database is MySQL5.7 to complete the construction

and support of the system database system. The

whole server is implemented by Django framework.

Through the introduction of the above key technical

theories, the overall environment of the system de-

velopment, the configuration of related software and

tools are determined, and the technical feasibility of

the overall project of the consumer buying behavior

analysis system is also clarified.

3 FUNCTION REALIZATION

3.1 Data Import

When the user enters the system, under the data

import function module, he selects the questionnaire

data prepared in advance and the purchase behavior

data of the enterprise's own e-commerce consumers

to upload to the system database.

3.2 Customer Classification and

Feature Selection

In this function module, the system will support

three functions: consumer classification, attribute

feature screening and purchase behavior rules min-

ing of consumers. In the user interface, select the

sample data, click on the consumer classification,

and the system will automatically call the RFM

ICPDI 2022 - International Conference on Public Management, Digital Economy and Internet Technology

384

model of K-mediod algorithm to complete the con-

sumer classification, and display the classification

results, as shown in Figure 3.

After obtaining the classification results, users

can see the corresponding chart contents under the

details page. Meanwhile, we will also see all the

attribute variables and 10 key attribute features con-

tained in this part of sample data, including prefer-

ential way, price, search time, communication initia-

tion time, collection number, comment number,

e-commerce platform functionality, social recom-

mendation, login mode, recommendation index, etc.,

will serve as an important basis for the prediction of

consumers' purchase behavior rules, and will be input

into the Decision Tree algorithm model for consumer

purchase behavior prediction.

3.3 Preference Relevance Analysis

According to the purchase behavior rules of different

types of consumers, select important value consum-

ers and general value consumers for data analysis,

and establish high-quality association rules from

three aspects: products, services and concerned

marketing methods. According to Apriori algorithm,

the correlation data of different categories of con-

sumers can be obtained, which provides scientific

data basis for the subsequent formulation of social

marketing strategies, the operation and sales of en-

terprise products and the improvement of service

methods. (Zhang, 2021)

4 CONCLUDING REMARKS

In this paper, based on the difficulties faced by the

social marketing strategy and decision-making pro-

cess of e-commerce platform in the era of digital

economy, an online interactive consumer buying

behavior analysis system with data mining technol-

ogy as the core and Web application technology as

the framework is proposed. The system can realize

user classification, feature prediction and correlation

analysis, find social marketing, product attributes,

service factors and other influence on consumer

purchase behavior, and take reasonable measures to

improve social marketing strategy, adjust the service

mode, rich service content, to maximize the interests

of the enterprise.

REFERENCES

Du Jiming. The analysis of Web development application

of Python language [J]. Digital Communication

World.2022.01

Liu Bingling. The Analysis and Research on the Influenc-

ing Factors of Social Media Marketing on Consumer

Behavior [J]. China-Arab States Science and Tech-

nology Forum.2021.07

Wang Yang, Jiang Xintong. The Design Research of MVC

Framework in Python and Django [J]. Computer and

Information Technology.2021.02

Zhang Peng. The Data Mining Technology and Practice in

the Age of Big Data [J]. China New Telecommunica-

tions.2021.11

Zhang Mengqi. The Analysis of Association Rules Based

on Apriori Algorithm [D]. Dalian University of Tech-

nology .2021.04

The Construction of Consumer Buying Behavior Analysis System Based on Data Mining in Social Marketing

385