Research on the Differences of Location Characteristics between Offline and Online Catering Spaces Under O2O E-Commerce

Abstract

The development of O2O (online to offline) e-commerce and instant delivery has not only made it easy to access information, but also it has overcome the limitation of walking distance, promoting the transformation of traditional commercial space and making online catering a common form of food consumption. Studying the spatial distribution of urban catering industry under new technologies is helpful to understand the current development of the catering industry, which is of great significance for guiding the development of the catering industry. However, the trend and mechanism of changes need to be further studied. Based on the spatial distribution of urban catering industry, this study establishes a framework for analysing offline and online catering spaces. In this framework, big data are used to conduct GIS kernel density analysis and spatial autocorrelation analysis to investigate the spatial distribution laws of the catering industry, and multinomial logistic regression analysis is applied to explore the main factors influencing location selection of offline and online catering spaces. The main results include the following: (a) the new pattern of catering space still follows the traditional location selection theory and (b) population density and road network density are the main factors affecting the catering space in Shenzhen.

Keywords

Offline catering online catering spatial density multinomial logistic regression location difference

Introduction

The popularity of O2O (online to offline) e-commerce has given birth to O2O internet companies, such as ‘Meituan’ and ‘ele.me’ in China, enhancing people’s access to information unprecedentedly. Online catering has become a common form of food consumption, which has a major impact on the operation of urban physical stores (He et al., 2019; Yang et al., 2020). As of 2018, the number of online catering users has reached 406 million, and the number of full-time food deliveryman has reached 9 million (IiMedia, 2019). At the same time, the rapid development of instant delivery in the logistics industry makes walking distance no longer the main factor limiting people’s consumption. In 2019, the number of instant delivery users in China reached 421 million, and the order volume exceeded 18.8 billion (Prospective Industry Research Institute, 2020). It can be seen that in the internet era, the continuous development and popularization of new pattern of catering space have brought new choices for people’s diet consumption. Therefore, studying the spatial distribution of urban catering industry under new technologies is conducive to understanding the current development of the catering industry, thus providing great significance for guiding the development of catering industry.

The development of e-commerce and instant delivery has promoted the transformation of traditional commercial space, but there is currently no feasible mechanism of research on the trend of changes, and the mechanism of changes needs to be further studied. Some scholars believe that residents in the area who have received high-quality education, have high consumption level and are young and familiar with the internet are more inclined to shop online, thus the spatial pattern of catering is susceptible to economic development, consumer scale, traffic accessibility and the impact of urban land prices. The online catering space concentrates on areas with high population density, convenient transportation and high consumption level (Calderwood & Freathy, 2014; Irawan & Wirza, 2015; Li et al., 2016; Wang et al., 2015; Xi et al., 2020). Some scholars believe that when people’s demands for stores are not easily available, they tend to shop online, and these cities are generally featured by low development level, underdeveloped real economy and inconvenient transportation (Boschma & Weltevreden, 2011; Cao, 2012; Farag et al., 2006). Even with the rapid development of ICTs, some scholars have proposed views such as Death of Distance and End of Geography, arguing that human activities, especially economic activities, are no longer constrained by traditional geographic space (Cao et al., 2012; Morgan, 2004). Therefore, it can be assumed that whether geographic space significantly constrains the development of e-commerce is a controversial issue. It can be seen the research on the characteristics of online catering space is complex. How to present the differences and influencing factors of offline and online catering spaces still requires further study.

With the changes in lifestyle, the traditional methods for spatial pattern research obviously cannot meet the requirements for modern research. For example, traditional survey data and statistical data are difficult to visualize the spatial pattern characteristics of the catering industry. As data mining technology is becoming increasingly mature, big data have brought new perspectives and opportunities to various academic fields, especially to the spatial research that requires geographic coordinate information and a large amount of catering data, so that we can use such data to estimate the overall status more accurately (Shi, 2018; Shirdastian et al., 2019; Tu et al., 2020; Wang, 2014). Meanwhile, the research methods have also gradually diversified and shifted from simple mathematical statistics and morphological description to the use of ArcGIS (Kernel Density analysis, Spatial Autocorrelation) and multinomial logistic regression and other methods. In this article, the combination of qualitative and quantitative methods is used to analyse the factors affecting the spatial distribution of the catering industry in Shenzhen. It is expected to provide references for the optimization of the urban catering pattern and the planning for the development of urban catering industry.

Based on the spatial distribution laws of the urban catering industry, this study uses Kernel Density analysis, Spatial Autocorrelation and multinomial logistic regression analysis to establish an analysis framework of offline and online catering spaces, exploring the distribution characteristics of offline and online catering spaces, and the relationship between the two patterns of catering spaces and population density, transportation and average house price. This study aims at the following: (a) exploring whether the offline and online catering spaces produce a location transformation under new technologies; (b) analysing the distribution of offline catering space and online catering space, and providing references for the physical business to adapt to the development of information technologies through transformation; (c) interpreting the changes in the influencing factors of catering space under the condition of new technologies, and providing a basis for the planning and design of catering space under the background of new technological progress.

Literature Review

This study focuses on the analysis of the differences in the characteristics and influencing factors of offline and online catering spaces. Therefore, the literature review of this study is mainly conducted from the following three aspects: (a) the differences in the operation modes between offline and online catering spaces, (b) the main influencing factors of catering space, (c) the research of catering space based on big data.

The difference in the operation modes between offline and online catering spaces. Offline catering space refers to the catering space that does not register any network services and requires consumers to consume in the stores. That is, the traditional catering space is the consumption process where consumers go to the restaurants to order, eat and pay. This consumption mode requires merchants to provide physical space to display and sell their services, and requires consumers to go to the restaurants to have face-to-face contact with the merchants. The payment method is mainly cash payment (Figure 1(a)). Online catering space includes the restaurants at the first stage of O2O e-commerce development: business group buying and those at the second stage: takeout service. The first stage of online catering O2O broke people’s information limitation in the catering space, creating group-buying catering, that is, consumers can select their favoured merchants through online apps, and then purchase the consumption coupons through electronic payment, and consume offline with the consumption coupons (Figure 1(b)); the second stage broke the constraints of walking distance, encouraging takeout services, that is, consumers can purchase services on the platform through electronic payment, and the platform and merchants employ deliveryman to deliver services, so that consumers and merchants can close the deal without going out (Figure 1(c)) (Liu et al., 2017; Yang et al., 2020; Zhang, 2017).

Figure 1.

Source: The authors.

The main factors influencing catering space. This article explores the main factors affecting the spatial development of O2O e-commerce from the perspective of location theory, which is the primary theory that explains the spatial distribution laws of economic activities in a marketized environment. The location theory of catering space mainly involves central place theory, innovation diffusion hypothesis and efficiency hypothesis and land tax theory. The main factors influencing catering space currently focus on indicators, such as regional economic level, population distribution, degree of transportation convenience, land price, residents’ purchasing power and differences between urban and rural areas (Chen et al., 2015; Prayag et al., 2012; Wu et al., 2019; Xi, et al., 2020; Xiao et al., 2013). This article attempts to study the factors that influence the catering space based on location theory. From the perspective of central place theory, it reflects the core area in central place that provides goods and services for consumers in nearby areas is also the area where a large population gathers. The level of the central place is highly consistent with the distribution density of population, so this article adopts population density to reflect the influence of central place theory on catering space. From the perspective of innovation diffusion hypothesis and efficiency hypothesis, due to the development of transportation and technological progress, urban residents can get services faster online, while rural residents are also more inclined to consume online due to the inconvenience of transportation. Obviously, the online catering space in different regions has different requirements for transportation. Therefore, the road network density can be used to express the impact of innovation diffusion hypothesis and efficiency hypothesis on the catering space (Boschma & Weltevreden, 2011; Farag et al., 2006). From the perspective of land tax theory, the closer the space to the city centre, the higher the income, the lower the freight, but the higher the rent, which will affect the intention of dealers to purchase, thereby influencing the aggregation pattern of catering space. Thus, land price can be used to reflect the impact of land tax theory on catering space.

Research on catering space based on big data. ‘Big data’, which were born in the 1980s, can be used to perform various comprehensive analyses to support the improved development of systems based on policies. The continuously evolving ‘big data’ science may be used to help scientists, policy makers and city planners to formulate policies, strategies, procedures and practices (Lan et al., 2018; Michael & Miller, 2013; Zhu et al., 2017). In the field of catering, big data can be combined with other new analysis methods. For example, the research by using the catering merchant data on Baidu map shows that Chengdu’s catering industry presents a ‘multi-core’ structure (Cheng, 2017); through the data on online group purchase, the aggregation and diffusion of Beijing’s catering industry space has been investigated (Wang, 2014). Traditional survey data and statistical data are difficult to visualize the spatial pattern of catering industry. The research on the location of urban catering space based on Baidu POI data can be combined with geographic coordinate information and a large amount of spatial data for spatial simulation analysis, so that the density of offline and online catering space can be better obtained, and can be used as the main basis of the dependent variable in the analysis of distribution characteristics and influencing factors of catering space in this article.

Methodology

Study Area

Shenzhen is one of the four major first-tier cities in China and the first special economic zone established in China. Since 1980, Shenzhen has developed from a small fishing village into an international metropolis, which is really rare in history. With its rapid development, new mode has emerged before the disappearance of old mode, which makes the coexistence of the new and old modes of catering spaces particularly prominent. Therefore, Shenzhen’s unique offline and online catering spaces meet the hypotheses of this study. The basic unit of this study is the street, which is a unique administrative division in mainland China, and has the same administrative status as township and town. There are 74 streets in the study area (Figure 2).

Figure 2.

Source: The authors.

Data Sources

The data for this study are mainly from ‘Meituan’ (https://SZ.meituan.com/) and ‘Anjuke’ (https://www.anjuke.com) (Meituan and Anjuke are currently China’s largest online food ordering platform and housing information platform). By crawling the information of Shenzhen’s online catering O2O merchants on ‘Meituan’, the information of a total of 53,713 merchants was crawled, among which a total of 33,215 merchants with the number of orders of 0 in this month were classified as offline catering space, and a total of 20,498 merchants with the number of orders of more than 0 in this month were classified as online catering space. Through the ‘Anjuke’ platform, the average house prices of 4,929 communities in Shenzhen were crawled, and then the land prices of various districts in Shenzhen were obtained (replaced by average house prices).

Research Methods

In this study, GIS kernel density analysis and spatial autocorrelation were used to analyse the distribution characteristics of offline and online catering spaces; multinomial logistic regression analysis was employed to explore the relationship between the two catering space patterns and population density, transportation and average house prices.

GIS Kernel Density Analysis and Spatial Autocorrelation

The kernel density estimation method can use the spatial attributes of the data sample itself to study the distribution characteristics of spatial data, and to study the distribution characteristics of points by detecting the spatial change of the point density in regular region (Monjarás-Vega et al., 2020; Wang et al., 2020). The kernel density estimation method is widely applied in the research of the spatial distribution characteristics of geographic elements. The quantitative study on the spatial distribution and aggregation of online O2O catering merchants and physical catering merchants in Shenzhen by using this method can be used as the main basis of the dependent variable in the analysis of distribution characteristics and influencing factors of catering space in this article. When kernel density method is used to calculate two-dimensional data, the value of d is 2, and the formula is expressed as follows:

K = \frac{1}{n h_{}^{2}_{π}^{}} \sum_{i = 1}^{n} k {[(1 - \frac{(x - x_{i}) + {(y - y_{i})}^{2}}{h^{2}})]}^{2},

(1)

where K is the kernel function; (x – x_i)² + (y – y_i)² is the distance between point (x_i, y_i) and (x, y); h denotes the bandwidth; n is the number of points in the range.

Spatial autocorrelation analysis refers to the spatial correlation between the attribute values of objects (or elements), or the correlation between attribute values is caused by the geographic sequence or geographic location of objects (or elements) (Builes-Jaramillo & Lotero, 2020; Xia et al., 2020). The spatial autocorrelation analysis in this article can be used for the analysis of difference between offline and online catering spaces, and can be used to analyse the correlation between influencing factors and catering spaces. Commonly used measures include Moran’s I, Geary’s C, Getis’s G coefficient, etc. In this article, Moran’s I coefficient is used as a research tool for spatial autocorrelation, and its calculation formula is as follows:

I = \frac{n}{\sum_{i = 1}^{n} \sum_{j = 1}^{n} W_{i j}} \times \frac{\sum_{i = 1}^{n} \sum_{j = 1}^{n} W_{i j} (x_{i} - \bar{x}) (x_{j} - \bar{x})}{\sum_{i = 1}^{n} {(x_{i} - \bar{x})}^{2}},

(2)

where n is the number of research objects; x_i, x_j represent the attribute value of the study area i and j respectively; $\overset{─}{x}$ is the mean of all attribute values in the sample; W_ij is the weight matrix that measures the relationship between spatial things, which is determined by spatial topology and distance. The value of Moran’s I is between ±1. A positive value indicates that the attribute values of the research objects are spatially positively correlated, a negative value indicates a negative correlation and a zero value indicates a random spatial distribution.

Multinomial Logistic Regression Analysis

In many cases, univariate linear regression may not be able to describe the results. Therefore, using the optimal combination of multiple independent variables to predict or estimate the dependent variable is more effective and more realistic than using only one independent variable for prediction or estimation (Xia et al., 2019). Through the multinomial logistic regression model, the relationship between the influencing factors and the density of catering space was established. The dependent variable is the density of offline and online catering spaces, and the independent variables (the main factors affecting the catering space have been discussed in the literature review) mainly include population density, road network density and land price. The multinomial logistic formula in this article is as follows:

y_{i} = β_{0} + β_{1} x_{1} + β_{2} x_{2} + β {}_{3}x_{3}^{} + μ_{i} .

(3)

Dependent variable y_i: density of offline catering space and online catering space, which is obtained by Baidu POI data and GIS Kernel Density analysis. The spatial density reflects the aggregation pattern of space in various districts of Shenzhen.

Independent variables: population density x₁: population density is the basic condition for the development of merchants, and is also the basis for the existence of catering space. This article selects the resident population density of the streets where the catering industry is located to describe population density. Road network density x₂: public transportation accessibility is an important factor affecting the location of business. Convenient public transportation can promote the development of businesses. The road network density of streets is used as an influencing factor of catering space. Land price x₃: land value will constrain the free choice of the location of business. In this article, the average house price of each street in Shenzhen is used to describe land price. β_i is the regression parameter and μ_i is the random error term.

Research Framework

In this study, whether the location selection of new catering space pattern is different from that of the traditional catering space is the research motivation, and methods such as kernel density analysis, spatial autocorrelation, multinomial logistic regression analysis were adopted to establish an analysis framework of offline and online catering spaces. The framework is mainly divided into two parts (Figure 3). The first part is the comparison between offline and online catering spaces. GIS kernel density analysis and spatial autocorrelation analysis were conducted on the information collected by big data, and the differences in the density and the number of space between offline and online catering spaces were compared according to the analysis results, then the change law of catering space pattern under the influence of new technologies was summarized in response to whether there is a location transformation of offline and online catering spaces under new technologies based on location theory. The second part is the analysis of the differences in the influencing factors between offline and online catering spaces. With the analysis result of spatial density in the first part as the dependent variable, and the population density, road network density and land price selected by location theory as independent variables, multivariate analysis of offline and online catering spaces was conducted. The analysis results can help us to grasp the change laws of the main factors influencing the location selection of online and physical stores, and also provide the basis for the planning and design of catering space under the impact of new technologies.

Figure 3.

Source: The authors.

Results Analysis

Data Analysis

The number, density and influencing factors of offline and online catering spaces were selected for correlation analysis (as shown in Table 1). The correlation coefficients between the number and density of offline and online catering spaces are 0.930 and 0.991, respectively, indicating a high degree of similarity in the spatial distribution and aggregation. The new pattern of catering space follows the traditional location theory. The correlation coefficients between the density of offline and online catering spaces and population density are 0.772 and 0.734, respectively, indicating a high significance. Although there is a certain similarity between the overall distribution of online catering space and that of offline catering space, offline catering space has a higher correlation with population density, suggesting that physical stores rely more on population density. The correlation coefficients between the density of offline and online catering spaces and road network density are 0.825 and 0.804, respectively, indicating that road network density has played an important role in both offline catering space and online catering space. But the impact of road network density on offline catering space is greater than that on online catering space. The correlation coefficients between the density of offline and online catering spaces and land price are 0.308 and 0.318, respectively, showing a low positive correlation. It can be seen that land price is not the primary factor affecting the location selection of catering space, but online catering space is more susceptible to changes in land price.

Table 1.

Autocorrelation between Elements of Offline and Online Catering Spaces

	Number of Offline Stores	Offline Store Density	Online Store Density
Number of online stores	0.930
Online store density		0.991
Population density		0.772	0.734
Road network density		0.825	0.804
Land price		0.308	0.318

Source: The authors.

Distribution Characteristics of Offline and Online Catering Spaces

There is a high degree of consistency between the density of offline and online catering spaces (Figures 4 and 5). It can also be seen from the correlation analysis above that the spatial characteristics are dominated by ‘dense distribution in the south and sparse distribution in the north, dense distribution in coastal areas and sparse distribution in inland areas’, and the ‘block’ distribution of online catering space is particular prominent. Therefore, the traditional spatial structure has not caused a fundamental change in the distribution of online catering space due to the emergence of new technologies. However, from a local perspective, some internal changes have taken place. It can be seen from the outline A that the aggregation of online catering space is significantly more obvious than that of offline catering space. The outline is mainly located in the Nanshan District of Shenzhen, which is the main industrial and business gathering area in Shenzhen, where a large number of labours and young people gather. According to the report of 2017 Chinese takeout service development, the user groups using online catering takeout platforms are mainly young people under 40 years old, occupying 99 per cent of the takeout service market. It can be seen that a large number of online restaurants are likely to gather in the emerging industrial area and business area. Outline B shows the opposite trend. The aggregation of online catering space is not as obvious as that of offline catering space. These areas are on the edge of the central area, where the population mobility is small, the economic activity is poor, and the demand for takeout service is relatively small.

Figure 4.

Source: The authors.

Figure 5.

Source: The authors.

In terms of the number of catering space, the number of online catering space stores does not exceed that of the offline catering space stores, revealing that offline catering space can provide higher quality services at the current stage. The percentage of the difference between the number of offline and online catering spaces of each street was compared. The formula is that the difference between the number of offline catering space stores and the number of online catering space stores is divided by the number of offline catering space stores. As shown in Figure 6, the bluer the colour, the greater the number of online catering space stores than the number of offline catering space stores, and the redder the colour, the opposite trend. In general, the number of offline catering space stores is greater than that of online catering space stores. Except for Shekou Street in Nanshan District, Dapeng Street in Dapeng New District, Nan’ao Street in Longgang District and Meisha Street in Yantian District, the number of offline catering space stores on all other streets is larger than that of online catering space stores. In the main urban areas, the number of online catering space stores has the tendency to exceed that of the offline catering space stores, indicating that the economic development can promote the development of online catering, while in the suburb, offline catering space still dominates, indicating that the online catering still has great potential for development in suburb.

Figure 6.

Source: The authors.

Analysis of the Influencing Factors of the Difference in the Spatial Distribution

In Data Analysis section, the univariate spatial relationship between the population density, transportation, land price and catering space density was analysed, but we know that the location selection of catering space is the result of the combined influence of various factors, so based on the above three factors, this article used multinomial logistic regression to analyse the location differences between online and offline catering spaces in Shenzhen.

As can be seen from Table 2, the significance between offline catering space and population density is 0.001, β is 0.326, the significance between road network density and offline catering space is 0.000, β is 0.618, indicating that road network density and population density have a great influence on offline catering space. The significance between land price and offline catering space is 0.215 and β is –0.09, indicating that land price has no effect on Shenzhen’s offline catering space or has a small negative correlation. Online catering space is highly similar to offline catering space in that population density and road network density have a great impact on the catering space, while land price has no influence on Shenzhen’s online catering space. It can be seen that population density and road network density are the main factors affecting the catering space in Shenzhen, but the impact of population density on offline catering space is greater than that on online catering space, and the influence of road network density on online catering space is greater than that on offline catering space. The results are contradictory to single-factor influence analysis, thereby confirming that the multi-factor influence analysis can reflect the internal checks and balances. The R² value of the regression model of offline catering space is 0.734, and the R² value of the multiple regression model of the online catering space is 0.687, suggesting that the three influencing factors have a better explanation for the total variation in the correlation with offline catering space, but there are still other important factors that affect the distribution of catering space density in Shenzhen.

Table 2.

Analysis Results of the Factors Affecting Offline and Online Catering Spaces

Independent Variable	Offline Catering Space			Online Catering Space
Independent Variable	Value of β	Distinctiveness	R²	Value of β	Distinctiveness	R²
Population density	0.326	0.001	0.734	0.281	0.008	0.687
Land price	–0.09	0.215		–0.085	0.277
Road network density	0.618	0		0.632	0

Source: The authors.

As shown in Table 3, after adding offline catering store density as the fourth factor to the predictor variables, the R² value of the regression model is 0.988, the degree of fit is high, the significance of the online store density is 0.000 and the β value has reached 1.068, indicating that the new factor analysis has had a significant impact on the online catering space. It further proves that the density of offline and online catering spaces is highly consistent.

Table 3.

Analysis Results of Adding Offline Catering Store Density as an Influencing Factor

Independent Variable	Online Catering Space
Independent Variable	Value of β	Distinctiveness	R²
Population density	–0.066	0.008	0.988
Land price	0.04	0.022
Road network density	–0.045	0.142
Offline store density store density	1.068	0

Source: The authors.

Discussion

The pattern of online catering space and that of offline catering space in Shenzhen are consistent, which can be attributed to the fact that although the rapid development of e-commerce enhances the ability of people to obtain catering information, the overall spatial structure and planning methods have not changed, resulting in no fundamental change in the distribution of online catering space. It also reflects that the distribution of physical restaurants has a significant impact on the online catering space.

Online catering space has high requirements for roads and population density. Areas with high road network density and population density are prone to produce space aggregation benefits. For example, Guangming District is the largest industrial cluster area in Shenzhen, with dense population of labour force. Based on China’s national conditions, most factories provide meals and accommodation, thus the demand for physical stores is not large. Secondly, this district is dominated by labour-intensive industry. The labour force is relatively cheap and young. Therefore, ordering meals has a great impact on price-sensitive people. The population density has a larger effect on the aggregation of offline catering space than that on the aggregation of online catering space, indicating that offline catering space needs to be in a smaller range to survive. Online catering space has a larger dependence on road network density than offline catering space. It can be seen that the range of instant delivery exceeding 3 km is the most important link connecting consumers and sellers, and is also a key factor determining the pattern of online catering space. The land price has almost no effect on the aggregation of catering space, indicating that the catering space is a consumer and transportation-oriented industry, but according to the results of big data, there is a high correlation between the aggregation effect of catering space and high-end blocks (the correlation coefficient is 0.605), indicating that the streets with high service levels are likely to produce the aggregation of catering industry.

The analysis framework of offline and online catering spaces constructed in this article can provide information support for government departments to formulate development strategies of future catering spaces, and to adjust city planning under the new technologies, and can also provide catering practitioners with a reference for future location selection. Besides, the analysis results can provide a basis for the location selection of industry, commerce, logistics, etc., because traditional industries are facing challenges brought by new technologies.

The innovations of this article are as follows: (a) The big data obtained from the internet platform cover a wide range, with high timeliness, which solves the problems such as the difficulty in sampling in the traditional study on urban spatial structure, poor accuracy and timeliness of data sample. (b) Exploring the change law of the location of offline and online catering spaces under the new technologies can help to discover the key factors that lead to changes, and is conducive to understanding the current development of the catering industry, providing important significance for guiding the development of the catering industry. (c) Challenging traditional location theory and exploring new laws of catering space under new technologies.

This study can be improved from the following aspects: (a) The influencing factors need to be further studied. There are many factors that affect the catering space pattern. From the perspective of the difficulty in obtaining information, the author discussed from three angles, which is relatively one-sided, and more influencing factors should be considered in the later research. (b) The spatial and temporal changes in the urban catering space pattern is the next research direction. With the development of O2O e-commerce, the catering space pattern has also changed. Especially, the online catering space has been developing for 7 years since 2013, undergoing the early stage, development stage and maturity stage. Therefore, the data can be tracked for a long time and space in future research, thereby making the research more complete. (c) New technologies are constantly emerging. This research on catering spaces is still relatively loose, mainly focusing on large-scale study in cities. Subsequent research can focus on communities to find specific changes in offline and online catering spaces, which can facilitate the practical operation.

Conclusion

The development of e-commerce and instant delivery has promoted the transformation of the traditional commercial space, and also spawned the ‘lazy economy’, which led to the rapid rise of online catering. This article conducted an in-depth analysis of the spatial layout of urban catering industry by establishing an analysis framework of offline and online catering spaces through methods such as kernel density analysis, spatial autocorrelation and multinomial logistic regression analysis. The main conclusions are as follows:

This article established a framework for analysing offline and online catering spaces, and used big data to conduct GIS kernel density analysis and spatial autocorrelation analysis, and compared offline and online catering spaces. Then, multinomial logistic regression analysis was employed to grasp the change laws of the main factors influencing location selection of online and physical stores.

The distribution and aggregation of offline and online catering spaces are highly similar, and the overall spatial pattern is characterized by ‘dense distribution in the south and sparse distribution in the north, dense distribution in coastal areas and sparse distribution in inland areas’, and ‘block’ distribution is prominent in the distribution of online catering space. It can be seen that the new form of catering space still follows the traditional location selection theory.

Population density and road network density are the main factors affecting the catering spaces in Shenzhen. The positive impact of population density on offline catering space density is greater than that on online catering space density, while the positive effect of road network density on online catering space density is greater than that on offline catering space density. This provides a basis for the planning and design of catering space under new technologies.

With the help of O2O e-commerce, network information has become the main approach for people to obtain social support and services, which exerts a huge impact on the development of catering industry. At the same time, the data capabilities of the O2O e-commerce platform have unprecedentedly enhanced the ability of people to obtain accurate and wide-range data. In addition, the complex realistic environment and the unpredictable results of the future lead to various problems in finding the distribution law of online catering space. Despite some conclusions, more efforts should be made in clarifying the complex realistic relationships.

Footnotes

Declaration of Conflicting Interests

The authors declared the following potential conflicts of interest with respect to the research, authorship and/or publication of this article: I confirm that I have included a citation for available data in my references section unless my article type is exempt.

Funding

The APC was funded by Youth project of National Social Science Foundation of China [21CSS001].

References

Boschma

, & Weltevreden

(2011). The evolutionary nature of b2c e-commerce in inner cities. http://citeseerx.ist.psu.edu/viewdoc/download?doi=10.1.1.195.619&rep=rep1&type=pdf

Builes-Jaramillo

, & Lotero

(2020). Closeness matters. Spatial autocorrelation and relationship between socioeconomic indices and distance to departmental Colombian capitals. Socio-Economic Planning Sciences, 70, 100662. 10.1016/j.seps.2018.10.013

Calderwood

, & Freathy

(2014). Consumer mobility in the Scottish isles: The impact of internet adoption upon retail travel patterns. Transportation Research Part A: Policy and Practice, 59, 192–203. 10.1016/j.tra.2013.11.012

Cao

X. J.

(2012). The relationships between e-shopping and store shopping in the shopping process of search goods. Transportation Research Part A: Policy and Practice, 46(7), 993–1002. 10.1016/j.tra.2012.04.007

Cao

X. J.

, Xu

, & Douma

(2012). The interactions between e-shopping and traditional in-store shopping: An application of structural equations model. Transportation, 39(5), 957–974. 10.1007/s11116-011-9376-3

Chen

, Liu

, & Liang

(2015). Characterizing the spatio-temporal evolution of retail business at transfer hubs of Guangzhou Metro. Acta Geographica Sinica, 70(6), 879–892. 10.11821/dlxb201506003

Cheng

(2017). Spatial structure and influencing factors of catering industry of Chengdu. Southwest Jiaotong University, China.

Farag

, Weltevreden

, van Rietbergen

, Dijst

, & van Oort

(2006). E-shopping in the Netherlands: Does geography matter? Environment and Planning B: Planning and Design, 33(1), 59–74. 10.1068/b31083

, Han

, Cheng

T. C. E.

, Fan

, & Dong

(2019). Evolutionary food quality and location strategies for restaurants in competitive online-to-offline food ordering and delivery markets: An agent-based approach. International Journal of Production Economics, 215, 61–72. 10.1016/j.ijpe.2018.05.008

10.

IiMedia. (2019). 2019 Chinese restaurant industry big data monitoring and new business model research report. IiMedia Life and Travel Industry Research Center.

11.

Irawan

M. Z.

, & Wirza

(2015). Understanding the effect of online shopping behavior on shopping travel demand through structural equation modeling. Journal of the Eastern Asia Society for Transportation Studies, 11, 614–625

12.

Lan

, Yu

, Xu

, & Wu

(2018). Temporal and spatial variation characteristics of catering facilities based on POI data: A case study within 5th Ring Road in Beijing. Procedia Computer Science, 131, 1260–1268. 10.1016/j.procs.2018.04.343

13.

, Zhang

Z.-B.

, & Wang

W.-J.

(2016). Spatial distribution and patterns and influential factors of large and medium supermarkets in Lanzhou. Economic Geography, 36, 85–93. 10.15957/j.cnki.jjdl.2016.09.012

14.

Liu

, Liu

, & Jiao

(2017). Research on the satisfaction of food takeaway in O2O mode. Business and Globalization, 5(3), 31–36. http://doi.org/10/12677/bglo.2017.53005

15.

Michael

, & Miller

K. W.

(2013). Big data: New opportunities and new challenges [Guest editors’ introduction]. Computer, 46(6), 22–24. 10.1109/MC.2013.196

16.

Monjarás-Vega

N. A.

, Briones-Herrera

C. I.

, Vega-Nieva

D. J.

, Calleros-Flores

, Corral-Rivas

J. J.

, López-Serrano

P. M.

, Pompa-García

, Rodríguez-Trejo

D. A.

, Carrillo-Parra

, González-Cabán

, Alvarado-Celestino

, & Jolly

W. M.

(2020). Predicting forest fire kernel density at multiple scales with geographically weighted regression in Mexico. Science of The Total Environment, 718(4), 137313. 10.1016/j.scitotenv.2020.137313

17.

Morgan

K. J.

(2004). The exaggerated death of geography: Localized learning, innovation and uneven development. Journal of Economic Geography, 4(1), 3–21. 10.1093/jeg/4.1.3

18.

Prayag

, Landré

, & Ryan

(2012). Restaurant location in Hamilton, New Zealand: Clustering patterns from 1996 to 2008. International Journal of Contemporary Hospitality Management, 24(3), 430–450. 10.1108/09596111211217897

19.

Prospective Industry Research Institute. (2020). Report of prospects and investment strategy planning analysis on China instant delivery industry (2020–2025). Prospective Industry Research Institute.

20.

Shi

(2018). The spatial pattern of takeaway O2O catering industry in Changchun and its interaction with physical restaurants. Northeast Normal University, China.

21.

Shirdastian

, Laroche

, & Richard

M.-O.

(2019). Using big data analytics to study brand authenticity sentiments: The case of Starbucks on Twitter. International Journal of Information Management, 48(3), 291–307. 10.1016/j.ijinfomgt.2017.09.007

22.

, Zhu

, Xia

, Zhou

, Lai

, Jiang

, & Li

(2020). Portraying the spatial dynamics of urban vibrancy using multisource urban big data. Computers, Environment and Urban Systems, 80, 101428. 10.1016/j.compenvurbsys.2019.101428

23.

Wang

(2014). Spatial distribution trend of urban service industries in the information times: a case study of Beijing’s catering industry. Urban Problems, 90–95. 10.13239/j.bjsshkxy.cswt.141116

24.

Wang

, Hao

, & Jiang

(2015). Locations and their determinants of large-scale commercial sites in Changchun, China. Acta Geographica Sinica, 70(6), 893–905. 10.11821/dlxb201506004

25.

Wang

, Ginzler

, & Waser

L. T.

(2020). Assessing structural changes at the forest edge using kernel density estimation. Forest Ecology and Management, 456, 117639. 10.1016/j.foreco.2019.117639

26.

, Nie

, Xu

, & Zhao

(2019). Distribution planning problem for a high-speed rail catering service considering time-varying demands and pedestrian congestion: A lot-sizing-based model and decomposition algorithm. Transportation Research Part E: Logistics and Transportation Review, 123, 61–89. 10.1016/j.tre.2019.01.003

27.

, Cao

, & Zhen

(2020). The impacts of same day delivery online shopping on local store shopping in Nanjing, China. Transportation Research Part A: Policy and Practice, 136(1), 35–47. 10.1016/j.tra.2020.03.030

28.

Xia

, Yeh

A. G.-O.

, & Zhang

(2020). Analyzing spatial relationships between urban land use intensity and urban vitality at street block level: A case study of five Chinese megacities. Landscape and Urban Planning, 193, 103669. 10.1016/j.landurbplan.2019.103669

29.

Xia

, Zhang

, Wang

, & Yeh

A. G. O.

(2019). Predicting the expansion of urban boundary using space syntax and multivariate regression model. Habitat International, 86(2), 126–134. 10.1016/j.habitatint.2019.03.001

30.

Xiao

, Chen

, Yuan

, & Cheng

(2013). Spatial pattern and location decision of chain supermarkets within large cities: A case study of Suguo supermarkets in Nanjing. Geographical Research, 32, 465–475

31.

Yang

, Gong

, Land

L. P. W.

, & Chesney

(2020). Understanding the effects of physical experience and information integration on consumer use of online to offline commerce. International Journal of Information Management, 51, 102046. 10.1016/j.ijinfomgt.2019.102046

32.

Zhang

(2017). Research on the O2O logistics distribution strategy and assessment of takeout catering. North China Electric Power University.

33.

Zhu

, Wu

, Li

, & Xiong

(2017). China’s regional natural resource allocation and utilization: A DEA-based approach in a big data environment. Journal of Cleaner Production, 142(Pt. 2), 809–818. 10.1016/j.jclepro.2016.02.100