Research on the Influencing Factors of Online Clothing Sales Based

on Binary Logit Regression

Yichen Gu

, Zejian Wang

and Yunfeng Xia

School of Economics and Management, Xidian University, Xian, 710000, China

Jurong Country Garden School, Jurong ,212446, China

Keywords: Data Analysis, Customer Satisfaction, Online Clothing Sales, Payment, Influence Factor.

Abstract: The main purpose of this study is to use the method of binary logit regression to conduct a comprehensive

discussion on "factors affecting online clothing sales". In the era of more and more developed online sales,

online clothing sales occupy a large part of online commodity sales, but at the same time, there are still many

problems in online clothing sales that need to be improved. The overall goal is to find the basic determinants

of customer satisfaction in online clothing sales, so as to give the factors for enterprises to increase online

clothing sales. In this study, some online questionnaires were first obtained to obtain the data of people

choosing online or offline shopping. Then, empirical analysis was conducted using the data set obtained from

Kaggle to identify the correlation of online fields that significantly affects product sales, and binary logit

regression analysis was carried out after processing these data. Through calculation, it is determined that the

payment method and the discount application has a significant positive impact on customer satisfaction, which

shows that it is essential for relevant managers to optimize these two aspects to promote the virtuous cycle of

online clothing sales.

1 INTRODUCTION

In 2022, the new driving force index of China's

economic development (100 in 2014) was 766.8, an

increase of 28.4 percent over the previous year, and

all sub-indicators have improved over the previous

year. Among them, the Internet economy index has

grown the fastest (Zheng, 2018). The internet

economy refers to the broad spectrum of economic

activities that rely on internet technology as a

platform, with the network serving as the medium and

application technology innovation at its core. The

present state of the Internet economy primarily

encompasses five main types: e-commerce, Internet

finance (ITFIN), instant messaging, search engines,

and online games. As a new economic form, the

Internet economy is highly innovative, which not

only shows the obvious technological sacrifice of The

Times, but also integrates the essence of the

traditional economy. The tangible economy

encompasses economic endeavors involving the

manufacturing and distribution of physical goods,

intangible products, and services. It comprises both

the material production and service sectors across

various industries, as well as the creation and

provision of intangible products, serving as a crucial

foundation for human sustenance and progress

(Zheng, 2018).

In recent years, China's online economy has

grown faster and faster, while the offline economy

has gradually lost its temperature. According to the

survey, by 2022, the national online shopping

replacement rate (the replacement rate of online

consumption to offline consumption) will reach

80.7% (Li & Shi, 2023). It can be seen that online

shopping has become the main way of residents'

consumption. With the gradual change of residents'

consumption patterns, an increasing number of brick-

and-mortar retail enterprises have started to undergo

transformation and enhancement. In this context, the

main sales channels at present are online sales on the

Internet platform and offline sales in traditional stores

(Yu, 2018). In addition, the domestic online game

industry was only in a stable growth trend in the first

half of 2017. As of June 2017, domestic online game

users have exceeded 420 million, 4.6 million more

than in 2016, accounting for 56.2% of the total

amount of domestic netizens (Gong, 2019).

In contrast, although China's real economy has

entered the 21st century, with the enhancement of

122

Gu, Y., Wang, Z. and Xia, Y.

Research on the Inﬂuencing Factors of Online Clothing Sales Based on Binary Logit Regression.

DOI: 10.5220/0012828900004547

Paper published under CC license (CC BY-NC-ND 4.0)

In Proceedings of the 1st International Conference on Data Science and Engineering (ICDSE 2024), pages 122-126

ISBN: 978-989-758-690-3

people's material living standards, China's domestic

physical retail industry has embarked on a phase of

swift advancement, experiencing an average growth

rate of around 10%, which showcases the increasing

purchasing power and consumer demand in the

market. Nevertheless, given the current

circumstances, the domestic physical retail industry

still struggles when compared to international

physical retail conglomerates in relation to overall

sales and operational scale. (Lei, 2016). In this case,

the real economy is not strong enough, and the real

economy has ushered in the impact of the Internet

economy. The rapid rise of the Internet economy has

made the traditional economy face unprecedented

challenges in terms of business philosophy, the

business model and the service model, which is

embodied in the loss of customer resources, the

decrease of deposit ratio and the decrease of

intermediate business volume (Yu, 2015).

But in fact, the fundamental goals of the

development of the Internet economy and the real

economy are the same. Whether it is the network

economy or real economy, the ultimate goal of its

development is to develop productivity (Cao, 2017).

Therefore, it is necessary to maintain the balance

between the Internet economy and the real economy.

From 2011 to 2018, China experienced a significant

surge in online retail sales, escalating from 0.78

trillion yuan to 9.01 trillion yuan, reflecting an

average annual growth rate of 43.6 percent. By 2018,

online retail sales in China constituted 10 percent of

the GDP. Meanwhile, offline retail sales saw a

notably lower average annual growth rate of only

4.2%, significantly beneath the average annual

growth rate of the GDP (Yu & Zou, 2020). However,

online shopping on the Internet and offline shopping

in the physical industry each has advantages and

disadvantages. Online shopping realizes cross-

domain transactions, so that consumers can buy the

goods they want without leaving their homes. On

shopping websites, consumers can easily understand

product information and make purchasing decisions.

Conversely, traditional offline shopping offers

consumers the chance to physically interact with

goods. Shoppers can meticulously examine products

and even engage in hands-on experiences with the

products themselves (Bai, 2019). As online-to-offline

integration becomes a new choice in the field of e-

commerce, the industry generally believes that the

"entry into the market" of online retailers will harm

the interests of offline retailers (Ding, 2019).

This study aims at analyzing the factors

influencing online sales in the fashion industry,

specifically identifying which types of clothing have

the greatest potential for online sales.

2 METHODS

2.1 Data Source

The study used data from China's National Bureau of

Statistics and the website Kaggle. The first dataset

collected people's choices for online and offline

shopping in 2022. The other dataset includes 14

factors that may affect people's online shopping

experience and related data, with a total sample size

of 3,900, which can fully illustrate the research

question.

2.2 Indicators and Analysis

The analysis has carefully chosen specific indicators

to deepen the understanding of the relationship

between online and offline shopping. These

indicators include product categories, prices,

purchase quantities, and the male-female ratio of

consumer groups. The analysis ensures that these

indicators will serve as effective tools for analyzing

and elucidating the complex dynamics of online and

offline shopping. In addition, targeted surveys have

been conducted to gain a deeper understanding of

consumers' attitudes towards online and offline

consumption. The survey aims at at providing a

detailed understanding of consumer perspectives for

research purposes, enabling a more comprehensive

analysis of the factors influencing consumers' choices

to purchase clothing online in both e-commerce and

traditional retail contexts (Table 1).

By leveraging, we sought to delve into the

complexities surrounding online and offline shopping

dynamics. While acknowledging the strengths of

these datasets, it is important to remain cognizant of

their limitations, particularly regarding temporal

aspects and the challenges associated with assessing

popularity. These considerations are essential for

maintaining the integrity and validity of the analyses.

The meticulous selection of indicators, combined

with the targeted consumer survey, forms the

cornerstone of the approach to unraveling the

multifaceted relationship between online and offline

shopping. With this comprehensive approach, the

goal is to provide valuable contributions to the current

knowledge base in this field, illuminating the

complex interplay of elements influencing consumer

behaviors within the spheres of e-commerce and

traditional retail.

Research on the Inﬂuencing Factors of Online Clothing Sales Based on Binary Logit Regression

123

Table 1: Delving into Metrics.

Indicator Mean ± standard deviation Variance Median Standard error

Age 44.068±15.208 231.271 44.000 0.244

Gender 1.680±0.467 0.218 2.000 0.007

Item Purchased 13.035±7.199 51.828 13.000 0.115

Category 2.002±0.897 0.804 2.000 0.014

Size 2.120±0.930 0.866 2.000 0.015

Color 13.109±7.222 52.151 13.000 0.116

Season 2.493±1.117 1.248 2.000 0.018

Review Rating 3.750±0.716 0.513 3.700 0.011

Subscription Status 1.270±0.444 0.197 1.000 0.007

Shipping Type 3.514±1.698 2.882 4.000 0.027

Discount Applied 1.430±0.495 0.245 1.000 0.008

Promo Code Used 1.430±0.495 0.245 1.000 0.008

Previous Purchases 25.352±14.447 208.719 25.000 0.231

Payment Method 3.512±1.691 2.858 3.000 0.027

2.3 Method Introduction

The study first conducted data screening, selecting

variables that may be related and analyzing the data

using binary logit regression. The Review Rating was

divided into two parts: 0 (dissatisfied) for ratings

between 2.5-3.5, and 1 (satisfied) for ratings between

3.5-5, which were used as the dependent variable y,

referred to as Customer Satisfaction. The study

selected Age, Gender, Category, Location, Purchase

Amount (USD), Size, Color, Previous Purchases,

Payment Method, Frequency of Purchases,

Subscription Status and Discount Applied as

independent variables. Descriptive analysis and

frequency analysis were performed on these variables

to highlight their characteristics and facilitate the final

binary logit regression analysis of Customer

Satisfaction.

3 RESULTS AND DISCUSSION

3.1 Basic Information

Using Age, Gender, Category, Location, Purchase

Amount (USD), Size, Color, Previous Purchases,

Payment Method, Frequency of Purchases,

Subscription Status and Discount Applied as

independent variables and Customer Satisfaction as

the dependent variable for binary logistic regression

analysis. The table above indicates the participation

of 3900 samples in the analysis, revealing the absence

of any missing data. (Table 2).

Table 2: Overview of Binary Logit Regression Analysis.

Name Options Frequency Percentage

Customer

Satisfaction

0 2106 54.00%

1 1794 46.00%

Total 3900 100.0%

Summary

valid 3900 100.00%

Hiatus 0 0.00%

Total 3900 100.0%

The model's fitting quality is assessed based on

the accuracy of its predictions. Table 3 indicates that

the overall predictive accuracy of the research model

stands at 82.74%, demonstrating an acceptable level

of model fitting. The prediction accuracy is 84.14%

and when the true value is 1 is 81.10%.

3.2 Model Results

The following table lists the data obtained by binary

logit regression analysis and the relevant results

obtained by these data (Table 4).

As can be seen from the above table, Age, Gender,

Category, Location, Purchase Amount (USD), Size,

Color, Previous Purchases, Payment Method,

Frequency of Purchases, Subscription Status and

Discount Applied are independent variables,

Customer Satisfaction is considered the dependent

variable for binary Logit regression analysis.

ICDSE 2024 - International Conference on Data Science and Engineering

124

Table 3: Summary of prediction accuracy with binary Logit regression.

Predicted Value

Forecast Accuracy Predicting Error Rate

0 1

true value

0 1772 334 84.14% 15.86%

1 339 1455 81.10% 18.90%

gather 82.74% 17.26%

Table 4: Summary of the results of the binary Logit regression analysis.

Sum

regression

coefficient

standard

error

z Wald χ2 p OR price 95% CI

Age 0.002 0.003 0.530 0.281 0.596 0.995 ~ 1.008

Gender 0.042 0.129 0.323 0.104 0.747 0.810 ~ 1.342

Category 0.024 0.054 0.456 0.208 0.649 0.922 ~ 1.139

Location 0.006 0.003 1.642 2.697 0.101 0.999 ~ 1.012

Purchase Amount

(

USD

)

-0.002 0.002 -0.879 0.773 0.379 0.994 ~ 1.002

Size -0.060 0.053 -1.138 1.294 0.255 0.849 ~ 1.045

Color 0.009 0.007 1.267 1.605 0.205 0.995 ~ 1.022

Purchases

0.004 0.003 1.313 1.725 0.189 0.998 ~ 1.011

Payment Method 0.061 0.029 2.121 4.497 0.034 1.005 ~ 1.123

Frequency of

Purchases

0.035 0.024 1.439 2.071 0.150 0.987 ~ 1.086

Subscription

Status

-0.218 0.154 -1.421 2.021 0.155 0.595 ~ 1.086

Discount Applied 0.372 0.155 2.399 5.757 0.016 1.071 ~ 1.966

intercept -14.877 0.575 -25.862 668.849 0.000 0.000 ~ 0.000

As can be seen from the above table 4, that means

Age, Gender, Category, Location, Purchase Amount

(USD), Size, Color, Previous Purchases, Payment

Method, Frequency of Purchases, Subscription

Status, Discount Applied can illustrate the 0.49

variation in Customer Satisfaction. It can be seen

from the above table that the formula of the model is:







−p



= −14.877 + 0.002 × Age + 0.042 ×

Gender + ⋯+ 0.372 × Discount Applied

(1)

The regression coefficient value of the Payment

Method is 0.061, and presents the significance level

of 0.05 (z = 2.121,p = 0.034 < 0.05) , which

means that the Payment Method will have a

significant benefit on Customer Satisfaction. And the

OR value is 1.062, which means that when the

Payment Method is increased by one unit, the change

(increase) of Customer Satisfaction is 1.062 times.

The regression coefficient value of Discount

Applied is 0.372, and presents the significance of

0.05 level (z = 2.399,p = 0.016 < 0.05). It means

that Discount Applied will have a significant positive

impact on Customer Satisfaction. And the OR value

is 1.451, meaning that when the Discount Applied is

increased by one unit, the increase in Customer

Satisfaction is 1.451 times.

3.3 Discussion

The summary analysis shows that Payment Method

and Discount Applied will significantly benefit

Customer Satisfaction. However, Age, Gender,

Category, Location, Purchase Amount (USD), Size,

Color, Previous Purchases, Frequency of Purchases,

Subscription Status does not affect Customer

Satisfaction. In a business environment, payment

methods and discounts are crucial factors that impact

customer satisfaction. Many retailers enhance

customer satisfaction by offering a wide range of

payment options and appealing discount policies. It is

also believed that customer satisfaction can be

influenced by various other factors, including age,

gender, category, location, purchase amount, size,

color, previous purchase history, purchase frequency,

Research on the Inﬂuencing Factors of Online Clothing Sales Based on Binary Logit Regression

125

and subscription status. Firstly, providing diverse

payment methods such as credit cards, debit cards,

Alipay or WeChat Pay can cater to customers'

different payment habits and needs while enhancing

shopping convenience and flexibility. Allowing

customers to select their preferred payment method

during the purchasing process will contribute to their

comfort and overall satisfaction. Additionally,

discount policies serve as effective promotional tools

that entice customers to make purchases thereby

increasing their desire to buy and overall satisfaction.

4 CONCLUSION

This study uses a binary logit regression model that

takes satisfaction as the dependent variable, using

age, gender, category, location, purchase amount

(USD), size, color, previous purchase, payment

method, purchase frequency, subscription status, and

discount application as the independent variables. At

the same time, this paper also takes into account some

control variables, such as purchase frequency, the

number of items in the shopping basket, etc. To

ensure the accuracy of the study results. The

relationship between many factors of online clothing

sales satisfaction is deeply discussed. By analyzing a

large number of sales data, this paper gets a series of

statistical results and calculates the extent to which all

independent variables affect consumer satisfaction.

When other potential variables are taken into account,

the study finds that Payment Method and Discount

Applied have a significant impact on consumers'

satisfaction with online clothing sales. Based on the

regression model of the research, some suggestions

for merchants' future sales can be made. When

formulating sales strategies, merchants should

seriously consider these two factors and improve

consumers' shopping experience through flexible use

of discount activities and targeted payment

promotion, so as to enhance sales performance. At the

same time, future research can further explore other

factors that may influence satisfaction to more fully

understand the market dynamics of online clothing

sales.

AUTHORS CONTRIBUTION

All the authors contributed equally and their names

were listed in alphabetical order.

REFERENCES

Y. Zheng, Rel. Inv. Entr. 8, 2 (2018).

B. Li and Y. X. Shi, Economic daily news, (2023).

H. Y. Yu, Journal of marketing research 1, 3 (2018).

Z. Y. Gong, Global Market 8, 17 (2019).

C. L. Lei, Business research 5, 3 (2016).

Y. M. Yu, Harbin Institute of Technology, (2015).

J. P. Cao, Management 1, 231 (2017).

W. T. Yu and M. Zou, Journal of UESTC 22(5), 10 (2020).

L. Bai, Journal of commercial age, (2019).

J. W. Ding, Business Economics Research, (2019).

ICDSE 2024 - International Conference on Data Science and Engineering

126