Analysis of the Distribution Characteristics and Influencing Factors of Advertising Billboards in Wuhan

The site selection of billboard locations has become crucial in promoting corporate image and expanding brand influence. Therefore, when choosing billboard placement locations, factors such as target audience and surrounding traffic conditions must be considered to fully leverage the value of billboard advertising. In this article, various analytical methods, and tools such as kernel density analysis, standard ellipse analysis, and geodetector are used to analyze the relationship between factors such as building density, road density, population density, and the number of shared bicycle rentals and returns per kilometer grid with the spatial distribution of advertising billboards in Wuhan City. Experimental results indicate that advertising billboards in Wuhan City exhibit spatial clustering, and the density of bus routes and stops, as well as the number of shared bicycle rentals and returns per kilometer grid, have a significant impact on their distribution.


INTRODUCTION
In today's market environment, the promotion of corporate image and the expansion of brand influence is an important task that every enterprise must face.To stand out in the competitive market, in addition to continuous innovation and optimization in advertising content, the choice of advertising location also becomes crucial.Therefore, reasonable billboard location selection has become an indispensable part of corporate image promotion strategy [1].When choosing the location of billboards, it is necessary to fully consider the brand image, audience groups, the surrounding advertising environment, as well as traffic conditions visibility, and other factors.Only by comprehensively considering these factors and making appropriate choices can we maximize the role of billboards in promoting corporate image and expanding the influence of the brand, thus obtaining better publicity effects and commercial value in the competitive market.
Currently, one of the common approaches to the billboard siting optimization problem is to use empirical judgments to determine the location of billboards, Wang et al. 2020 used a combination of machine learning and empirical methods to study the siting problem of digital signage [2].They proposed a siting method by training on digital signage point data and modeling factors to improve interpretability and computational efficiency.In the field of siting, there exists a variety of other methods applied in addition to empirical methods [3,4].These methods include but are not limited to, the use of multi-source data in conjunction with GIS tools to determine optimal placement locations, the use of decision support systems for optimization of siting, and the use of algorithms for optimization of siting [5,6].
show that the proposed method has higher precision and recall in location recommendation, which leads to a better recommendation effect and can further improve the layout of digital signage.Sadeghi et al. 2019 investigated how to use ArcGIS to select the optimal installation locations for banners and billboards in health promotion campaigns [8].Zhang et al. 2019 combined multi-source data based on GIS software to study the spatial distribution characteristics of outdoor commercial digital billboards within the Sixth Ring Road in Beijing [9].The study used point pattern analysis, spatial clustering analysis, and correlation analysis methods.It reveals the aggregation of billboard distribution and divides it into three categories: traffic-oriented, population-oriented, and market-oriented, and specifies the type of multi-source data with the highest correlation, to guide the optimal location selection of billboards.Liu et al. 2019 proposed an interactive visualization and analysis system called SmartAdP for large-scale cab trajectory data in selecting the location of billboards in visual analytics [10].The system integrates a new application-driven mining model and several well-designed visualization and interaction techniques, which contain a location view and ranking view, support comparing different scenarios from multiple perspectives, and demonstrate the usefulness and effectiveness of the system through case studies and expert interviews, and the methodology can be generalized to other location selection problems.Wakil et al.2021 proposed a spatial Decision Support System (SDSS) that aims to mitigate the urban visual pollution caused by billboards by balancing the needs of various stakeholders and helping users to identify potential locations for new billboards [11].The system utilizes geospatial open-source technology and hierarchical analysis of hierarchy (AHP) methods to support spatial decision-making, and it is functional in identifying hotspots and suitable locations for billboards.
The optimization of billboard placement is greatly facilitated by an understanding of the current spatial distribution of existing billboards.This understanding can provide valuable insights into the factors that should be considered in the site selection process.Traditional methods relying on subjective memory and experience are prone to distortion over time, making them less reliable.However, obtaining specific and comprehensive analytical results can be time-consuming and resource-intensive.The development of spatial analysis methods has effectively addressed this issue.
In this study, we employ spatial analysis methods such as standard deviation ellipses and geographic detectors to investigate the spatial distribution of billboards in Wuhan, as well as the factors influencing their distribution.This research aims to provide relevant insights for optimizing billboard placement.

Study Area
Wuhan is located in the central part of China, on the northern bank of the Yangtze River.It borders Huangshi to the east and is surrounded by Hubei Province on all sides.Its central coordinates are approximately 114.31°E longitude and 30.52°N latitude.As the provincial capital of Hubei, Wuhan holds significant political, economic, cultural, and historical importance for both the province and China.
Wuhan serves as the political center of Hubei and is one of the key political, cultural, and economic hubs in the province.In 2022, Wuhan's GDP reached 1.89 trillion yuan, ranking it first in Hubei and eighth in the nation.The city houses governmental institutions, diplomatic missions, and official media outlets for Hubei Province, making it a decision-making center for political events and major activities.Wuhan's significance extends beyond Hubei and exerts a wide-reaching influence at the national level.This article focuses on the entire city of Wuhan and analyzes the spatial distribution of billboards throughout the city.It also discusses subsequent billboard location selection work.Wuhan's ring road is one of the key transportation arteries connecting the city center with its suburbs.With the city's rapid development, the ring road has gradually become a significant symbol of urban development, playing a crucial role in various aspects of the economy, culture, and society.Along the expressway, numerous important enterprises, commercial institutions, educational and cultural organizations, and medical facilities are concentrated, along with a substantial population and material resources.Research on the distribution of billboards throughout Wuhan can help understand the strategic layout of billboards, improve the city's promotional effectiveness, and contribute to the city's development and prosperity.Selecting suitable billboard locations can better meet the needs of citizens, businesses, and the government, while also maintaining the city's aesthetics and environmental quality.

Data
The data in the study area covers the following: building dataset, road dataset, population dataset, public transportation route dataset, bike share service point dataset, charging station point dataset, and billboard point dataset.These data can be categorized into four main categories of influences, which are geographic influences, footfall influences, audience influences, and mobility influences.In this, the building data is sourced from the National Tibetan Plateau Data Center, which provides accurate latitude and longitude information of the building.These data serve as the basis for determining the visibility of billboard placement, helping identify the optimal locations for billboards.Road data, on the other hand, contains topological information about roads and road types, further helping us to assess the accessibility of billboards and their location in the road network.Bike share service points and charging station points are often strategically located and geographically fixed in the city, and this data provides us with crucial information about the range of users' activities.All of these data are categorized as geographic influencers because they are directly related to geographic location.Demographic data, on the other hand, reflects the distribution of the population as well as the activity level of the area, which in turn are directly relevant for determining the radius of the billboard.Therefore, demographic data is part of the foot traffic influencing factor.Billboard spot data reveals the distribution and density of billboards in different geographic areas, which is used to build audience profiles and help realize more accurate advertising.Therefore, billboard spot data is categorized as an audience-influencing factor.Public transportation route data can reflect the mobility of vehicles, which has a certain impact on the degree of advertising exposure.
All these data have a positive effect on the location of billboards, and the comprehensive use of these data to select the location can improve the exposure rate of billboards so that the advertising effect can be as optimal as possible.

Workflow
The research framework of this article is shown in Figure 1.In order to explore the characteristics of billboard distribution in Wuhan, this study employs various spatial analysis methods.These methods include the Average Nearest Neighbor method and Kernel Density analysis to investigate whether billboards exhibit clustered distribution and the extent of clustering.Additionally, the Standard Deviation Ellipse analysis is used to examine the directional distribution characteristics of billboards.Hotspot and Coldspot analysis is used to identify areas with high and low concentrations of billboards.Furthermore, Geodetector analysis is employed to explore the factors influencing billboard distribution.
First, basic data processing was carried out to gain a preliminary understanding of the spatial distribution of billboards in Wuhan.Subsequently, various methods such as the Average Nearest Neighbor were employed to provide a detailed description of the spatial distribution characteristics of billboards in Wuhan.The experimental results obtained will contribute to determining suitable weightings for factors influencing the optimization of billboard placement in Wuhan.

Standard Deviational Ellipse Analysis
The standard deviation ellipse uses parameters such as the center, major axis, minor axis, and orientation angle to quantitatively describe the spatial distribution of the study object.It precisely reveals various characteristics of economic spatial distribution and analyzes the spatial distribution characteristics of discrete datasets.
It offers a global perspective on the spatial distribution of geographic elements, explaining their centrality, dispersion, and orientation over time and space.
The center point of the ellipse represents the weighted center position of the dataset, the major axis indicates the direction of data distribution, and the minor axis represents the extent of data distribution.A larger minor axis indicates greater data dispersion, while a higher eccentricity signifies a more pronounced directional aspect.The formulas for calculating these fundamental parameters are as follows.Standard Deviation Ellipse Center: In equation ( 1),  ! and  !represent the coordinates of the standard deviation ellipse center. & is the weight of the th pixel, and in this article, the weight corresponds to the nighttime light brightness value of the pixel. # and  # are the coordinates of the th pixel, and  is the total number of pixels.Ellipse Orientation Angle： In equation ( 2),  represents the orientation angle, while  # and  # are the differences between the coordinates of the ellipse center and the coordinates of the th pixel.Major and Minor Axis Standard Deviations: In equation ( 3),  ' and  ) represent the standard deviations along the  -axis and  -axis, respectively, which correspond to the lengths of the major and minor axes of the ellipse.[12]

Average Nearest Neighbor
Average nearest neighbor is a spatial statistical analysis method used to measure the degree of clustering in a spatial pattern.Specifically, the average nearest neighbor method calculates the distances from each feature point to its nearest neighboring feature point and then computes the average of these distances.It then compares this average distance with the average distance that would be expected under a completely random distribution.If the actual distance is smaller than the expected distance, it indicates a tendency for feature points to cluster; if the actual distance is greater than the expected distance, it suggests a tendency for feature points to be dispersed.
The nearest neighbor index is used to measure the clustering of billboards in Wuhan, Hubei Province, and to characterize the spatial distribution pattern of billboards in the city.Its calculation formula is as follows: In equation ( 4),  represents the nearest neighbor index,  % is the average distance between nearest neighbor points,  * is the expected nearest neighbor distance in a random distribution, and  is the point density.When  = 1 , it indicates that the spatial distribution of billboard points in Wuhan is random, and it represents the value of the average distance  % between nearest neighbor points, while  * represents the average distance between nearest neighbors in a random distribution pattern.When  > 1, it suggests that billboard points in Wuhan tend to be uniformly distributed, and when  < 1, it indicates that billboard points in Wuhan tend to be clustered.
The average nearest neighbor method is highly useful in various fields such as studying geographical spatial patterns, ecological distributions, and social spatial structures.It helps us understand whether feature points in a specific area exhibit spatial correlations and clustering phenomena.[13]

Kernel Density Analysis
The basic idea behind Kernel Density Analysis is to treat each data point as the center of a kernel function and then form a local region around each point.By taking a weighted average of the data points within these local regions, density estimation values are calculated for each location.The weights used for the weighted average are typically based on a distance-based function, such as a Gaussian kernel function.
Using Kernel Density Analysis allows for an intuitive representation of the spatial distribution of billboard points in continuous areas in Wuhan.This indirectly reflects the clustering of the study target in space.The calculation formula is as follows: In equation ( 5),  # represents the kernel density at any point  in space,  + is the weight for the research object ,  #+ is the distance between spatial point  and research object  M #+ < 0O, R is the bandwidth of the selected rule area, and  is the number of research objects  within the bandwidth  range.The calculation is performed for any point in space and its surrounding area.It helps analyze the spatial distribution patterns and characteristics of observed objects.Data points closer to the center point are assigned higher weights, while those farther away are assigned lower weights.The estimated density for each point is a weighted average density of all points in that area.
Kernel Density Analysis is a powerful data analysis method that involves smoothing discrete data and estimating probability density distributions.It helps us understand and describe the distribution patterns and trends of data on a specific variable.[14]

Geodetector
Geodetector is a statistical method used to detect spatial variations and underlying driving factors.It is widely applied in geographical research to investigate various influencing mechanisms.The fundamental concept behind Geographic Detector is that if an independent variable has a significant impact on a dependent variable, then the independent variable and the dependent variable should exhibit similar spatial distributions [15,16].Geographic Detector includes four categories of detectors: differentiation and factor detector, ecological detector, interaction detector, and risk detector.
In this study, the Geographic Detector model is used to quantify the individual and interactive effects of various influencing factors on the spatial distribution of billboards in Wuhan City.It aims to reveal the primary influencing factors of billboard spatial distribution in Wuhan and the ways in which these factors interact.
As shown in Table 2, this study selects eight independent variables, including population density, bus stop density, bus route density, road density, charging station density, building density, the number of shared bicycles borrowed per kilometer grid, and the number of shared bicycles returned per kilometer grid.This article primarily utilizes the Differentiation and Factor Detector in the Geographic Detector to study the explanatory power of various influencing factors on the spatial distribution of billboards in Wuhan City.It employs the Interaction Detector to assess the interaction effects between two influencing factors on the types and strengths of their impact on the spatial distribution of billboards in Wuhan City.
The formula for calculating the -value is as follows: Where ℎ = 1, … ,  represents the strata of the influencing factor ;  , ( and  ( are the variances of stratum ℎ and the overall area, respectively;  , and  are the sample sizes of stratum ℎ and the total sample size, respectively.The -value falls within the range of [0,1], indicating the magnitude of the impact of this influencing factor on the spatial distribution of billboards in Wuhan City.A larger value indicates a greater impact.

Spatial Analysis
By processing and converting the data records within the table that represents the billboard data of Wuhan City into vectorized points, and subsequently visualizing them by superimposing these points onto a foundational map of Wuhan City, we have successfully generated a spatial distribution map (Figure 2).Upon examining the spatial distribution map of billboards in Wuhan City, it becomes evident that billboard locations are notably concentrated within the central urban regions of Wuhan City.This concentration includes districts such as Jianghan District, Hanyang District, and Wuchang District.This observation aligns seamlessly with the primary areas of urban development within Wuhan City, specifically the convergence point of the Hanjiang and Yangtze Rivers.These regions are known for their robust economic development and consequently exhibit a higher demand for billboards.Building upon this insight, a comprehensive series of spatial statistical analyses has been carried out concerning the distribution of billboards within Wuhan City.

Standard Deviational Ellipse
The creation of standard deviational ellipses offers valuable insights into the spatial distribution characteristics of geographic features, shedding light on their centrality, dispersion, and directional trends.Within ArcMap, the modification of the Ellipse Size parameter allows us to generate error ellipses with varying coverage rates, all based on point information, as exemplified in Figure 3.The error ellipses illustrated in this study encompass the majority of billboard locations, effectively encapsulating approximately 98% of the billboard point data.The computation of error ellipses yields essential information, including the central coordinates of the ellipse area, the standard distance of the major axis, the standard distance of the minor axis, and the orientation of the ellipse.Specific parameters can be found in Table 3 for further reference.The coordinates and distances in the table are originally in Mercator projection.After conversion, it is determined that an error ellipse, centered at approximately 30°34'40.74"N,114°18'21.81"E,can effectively encompass around 98% of the billboard locations within Wuhan City.Moreover, it becomes apparent that the clustered regions of billboards predominantly follow a northwest to southeast distribution pattern, covering a substantial portion of the central urban development zone in Wuhan City.This distribution corresponds to the prevailing economic development direction of Wuhan City.Notably, the eccentricity and angle of the ellipse signify a significant directional trend in the distribution of billboard locations.
The analysis of error ellipses effectively covers the majority of billboard locations in Wuhan City.Subsequent spatial statistical analysis will build upon this foundation to obtain more meaningful results.

Average Nearest Neighbor
The spatial pattern among the points is determined by comparing the average distance of the nearest point pairs.As depicted in Figure 4, the analysis of average nearest neighbor values reveals a clustering trend, which is further supported by the normal distribution curve in the analysis report, showing a p-value less than 0.05.This finding suggests that billboard locations are not randomly distributed, as also highlighted in the report.Notably, the Z-score in the report is a very small negative value of -97.633471, reinforcing the conclusion that the distribution of billboards in Wuhan City exhibits significant clustering and is not a result of random data generation.In the report, the observed average distance is determined by computing the average distance between each billboard location's centroid and that of its nearest neighbor.This calculation results in an average distance of approximately 137.9687 meters.On the other hand, the expected average distance is obtained by dividing the total number of points in the area by the size of the area, yielding a value of about 725.6240 meters.The nearest neighbor ratio is then derived as the ratio of the observed average distance to the expected average distance, which equates to 0.190125.This calculation further substantiates the clustering of billboard locations in Wuhan City.

Hot Spot Analysis (Getis-Ord Gi*)
The statistical model employed in this paper is the Getis-Ord Gi* model.Upon completing the data processing, the attribute table will feature a column named "Gi_Bin," with values spanning from -3 to 3.This column plays a crucial role in identifying statistically significant hotspots and cold spots.As illustrated in Figure 5, values within the range of -3 to 3 indicate 99% statistical significance, -2 to 2 indicate 95% statistical significance, -1 to 1 indicate 90% statistical significance, and 0 signifies no statistical significance.

Kernel Density Analysis
In this experiment, the study area encompasses the expansive expanse of Wuhan City, with a relatively dense distribution of billboard locations.Consequently, numerous regions register a value of 0. During the experimental analysis, these areas devoid of value are deliberately masked out to eliminate their influence.
As illustrated in Figure 6, the results of the kernel density analysis manifest distinctive multi-peaked patterns, indicative of polarization.The peak regions prominently converge in the Jianghan District, renowned for its commercial district status and the renowned pedestrian street, Jianghan Road.This suggests a heightened demand for billboards within bustling commercial pedestrian zones.Additional peak regions are also strategically positioned near shopping malls and educational institutions, all of which necessitate a substantial number of billboards for revenue generation.Based on the array of spatial statistical analyses conducted above, we can discern the spatial distribution patterns of billboards in Wuhan City.The distribution is characterized by clustering, with a primary concentration in the central urban areas, including Wuchang, Jianghan, and Hankou.Notably, this distribution closely aligns with the locations of commercial districts and pedestrian streets.The insights gleaned from this spatial analysis of billboards prove invaluable for making informed decisions regarding the optimal selection of potential billboard locations.Furthermore, it forms a solid foundation for algorithmic optimization in the process of site selection.

Geodetector Results
To facilitate better calculations, the data mentioned in section 2.2 was processed into 1-kilometer grid cells, ensuring that the experiments were conducted at a consistent geographical spatial scale.Through the Differentiation and Factor Detector tool of the Geographic Detector model, we analyzed the degree of influence of different influencing factors on the spatial distribution of billboards in Wuhan City.We define spatial heterogeneity as the differences in billboard distribution density per kilometer grid in Wuhan City.The results of the factor detector show that these influencing factors all significantly explain the spatial distribution of billboards in Wuhan City (Table 4).The factor detector results, sorted by their q-values, are as follows:

CONCLUSION
This study primarily focuses on the spatial distribution of billboards, bus stations, charging stations, bus routes, road networks, buildings, bike-sharing stations, and population data within the Wuhan city area.By utilizing various spatial analysis methods such as kernel density analysis, spatial correlation analysis, and hotspot analysis, we explore the influence of different factors on the spatial distribution of billboards.From the aforementioned analysis, it is evident that the optimal placement of billboards should be closely associated with public transportation infrastructure and other relevant structures.This association is beneficial in enhancing the visibility and coverage of advertisements, which, in turn, not only increases the effectiveness of advertisements but also serves as a means of communication, providing information and navigation services.More details of the project are available at: https://github.com/HIGISX/hispot.
It is worth noting that the methods and tools used in this study are rooted in GIS (Geographic Information Systems) technology, highlighting the crucial role of GIS in urban planning and management.The widespread application of GIS technology allows us to better understand the spatial distribution characteristics of cities and provides a scientific basis for urban planning and management.Furthermore, the methods and tools employed in this paper can serve as a reference for spatial distribution analysis in other cities.However, this study has certain limitations, such as the failure to consider the impact of competition between different types of billboards on their placement.In the future, we will continue to improve and expand spatial analysis methods, focusing on delving deeper into the spatial distribution patterns of billboards in Wuhan.This will provide more valuable insights for optimizing the placement of billboards.
Overall, this research provides valuable insights into revealing urban spatial distribution patterns.It emphasizes the importance of combining analytical methods with real-world considerations when conducting spatial analysis, ultimately contributing to making informed decisions in urban planning and management.

Figure 1 :
Figure 1: The Research Framework of This Article.

Figure 2 :
Figure 2: Spatial Distribution of Billboards in Wuhan City

Figure 4 :
Figure 4: The report on the average nearest neighbor results

Table 2 .
Independent Variable Data Summary

Table 3 .
Error Ellipse Parameters Thus, the factor explaining Bus Route Density has the highest explanatory power, indicating that the spatial distribution of billboards in Wuhan City is most strongly influenced by the distribution of bus route density, meaning that there is a strong consistency between the bus route density and the spatial distribution of billboards in Wuhan City.Secondary factors are the Number of Shared Bicycle Returns and the Number of Shared Bicycle Rentals per Kilometer Grid, indicating that shared bicycles are also important factors affecting the spatial distribution of billboards in Wuhan City.

Table 4 .
Factor Detector Results X6) and Bus Route Density (X3), Number of Shared Bicycle Returns per Kilometer Grid (X7) and Bus Stop Density (X2), and Number of Shared Bicycle Rentals per Kilometer Grid (X8) and Bus Stop Density (X2) are relatively large.The interactions between Road Density (X4) and Population Density (X1) and between Building Density (X6) and Population Density (X1) are relatively small.