The Construction of Consumer Buying Behavior Analysis System
Based on Data Mining in Social Marketing
Di Zhang
1,2
, Dongphil Chun
1
and Minghao Huang
3*
1
Pukyong National University, Graduate school of Management of Technology, Busan, 48548, Korea
2
Economic and Management School, Hulunbuir University, Inner Mongolia Hulunbuir 021008, China
3
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
Copyright
c
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