1 Introduction

Cities serve as the hubs of industries and business (Bélissent, 2010). Some cities were undergoing rapid growth in the leisure and entertainment industry in recent decades and transformed from centers of production to centers of consumption (Maitland & Newman, 2014). The presence of agglomeration in service provision in cities is not only attractive to locals but also to the population of the hinterland (Fischer, 2010; Lichter & Brown, 2011). For example, promising wealth and employment opportunities drew massive migrations from the hinterland to cities (Lichter & Ziliak, 2017). However, despite the ample amount of literature on the new roles of cities in the service-driven economy, many questions remain concerning how various sectors of the economy interact in urban settings and how the formation of a new economic sector is influenced by services of other sectors. The live music industry is one experiencing significant growth effects according to its participants (Fay, 2014; Green & Bennett, 2019). When examining the factors influencing the evolution of live music, a greater emphasis is placed on the social and cultural atmosphere within cities and the demands of residents for such forms of entertainment (Brown & Knox, 2017). However, the relationship between live music and its related expenses at the regional or city levels has received less research attention, and little is known, for example, about how transportation and meals for events contribute to the evolution of the live music industry.

Discussions about establishing culture-led strategic planning for cities and regions that contribute to the place competitiveness and reputation of the city have led to an exhaustive number of works. Ferilli et al., (2017) have highlighted the employment opportunities that cultural and economic regeneration provides. The evidence from empirical research investigating the role of art festivals in local economic and cultural development suggests that investments in the cultural industry create jobs and generate regional economic growth (Comunian, 2017). Music events as a part of arts festivals provide new insight into regional development plans to help increase city competitiveness and advance culture-led economies (Behr et al., 2020a, 2020b). Kruger and Saayman (2017) have emphasized the economic potential of music festivals, which were carried out to benefit local economies. The music city is a typical example where cultural and creative industries flourished (Florida et al., 2010). For example, Nashville has integrated its music scene into ad campaigns for travel promotional videos to drive tourism to the city (Rosenzweig et al., 2018). However, literature on the emergency of music cities in the U.S. remains scarce (Wynn & Dominguez-Villegas, 2016).

Country music is a unique American art form that originated mainly in the American South and was rooted in working-class Americans (Abramson, 2002). It is one of the most significant cultural items unique to the American South (Carney, 1980). Various factors from social, economic, and political aspects have helped to shape the modern country music scene (Martinez, 2020). Within the field of country music studies, a great deal of research has been conducted on music circulation and transmission, with great attention paid to the social and cultural landscape changes therein. For example, Feder (2006) addressed country music evolution by examining the history and culture of American South communities. Geographical surveys of music in urban areas are mainly part of broader studies of urban culture and have been incorporated into the topics of urban development. Yet, geographic factors are often overshadowed by the music’s cultural significance. Studies rarely assess geographic variations of music events and the effect of socioeconomic factors on these events. Identifying factors that have an impact on certain localities of country music can help inform musicians and policymakers.

2 Literature review

2.1 Music in geographical analysis

The music industry has evolved over the past few decades and relatively little work has yet been done in music geography compared to studies of other cultural industries (Connell & Gibson, 2004). Early geographical studies of music shared an emphasis on musical cartography. As outlined by Kong (1995) that the geographical study of music was driven by descriptive and empirical approaches traditionally. Mellander et al., (2018) used data compiled from an online survey to map music preferences. Song lyrics were decoded to reveal geographical meanings of music to reveal urban landscapes across space and time (Keeling, 2011). Some data-driven approaches were also conducted to map genre-specific music scenes (Cohen, 2012). The rapid change of geographical context requires the study of music associated with specific places in a more considerable depth. However, despite there are many attempts to define the geographic context of music from multiple aspects such as social geography, economic geography, and demographic studies, less attention has been paid to the implication of localized understandings and spatial awareness of live music.

Music has only in the past few decades become a topic of interest to cultural geographers. Music spatiality is most often figured through conceptions of culture geography (Revill, 2000). By focusing on music as an essential element of culture, studies were conducted to understand the broader geographical conditions from which distinctive forms of musical expression emerged. Many argued that, beyond the study of music itself, there is a need to critically consider music and space (Holt, 2010; Hudson, 2006). Lorenzen and Frederiksen (2008) stated the importance of the perception of geographical processes in understanding how localization facilitates music production and consumption. Different spatialities were identified as a mirror of the emergence of diversity within and across cultural industries (Pratt, 2004). Way et al., (2019) utilized place-based methodologies and musical taste profiles as lenses to understand the influences that geographic relocation of individuals had on the evolution of music. Since music often reflects the spatial dynamics of socioeconomic systems through listener and artist activities, it is worthwhile to examine possible geographical links between musical activities and their decisive factors.

Interests within geography in the spatiality of musical activities can be traced to attempts of exploring spatially dynamic elements of musical expression. The geography of music can be learned through multiple music features, such as music production, music distribution, and music consumption. Verboord and Noord (2016) examined music distribution and found that music-producing activities are more highly populated in central cities than in peripheral locations. The music industry network was also analyzed through an examination of place-specific attributes of a city (Cummins‐Russell & Rantisi, 2012). Kruse (2010) took a commercialization perspective and found music festivals and club concert touring as means by which musicians can connect with fans across geographical boundaries. By investigating spatial dynamics of musicians and music firms in the U.S. over 30 years, Florida et al (2010) noted that the geographically concentrated musicians and music assets allow growth of multimedia production and increase of economic outcomes.

Music plays diverse roles in economic, cultural, and political aspects of society. The geographical importance of music lies in its role in reflecting and shaping a certain process across space accordingly. Geographical processes can be identified in the study of music. An example is Watson (2008)’s examination of knowledge transfer in the music industry, where they explained the importance of geographical scales in their models. Haverluk (1997) used music that was popular on radio stations to analyze the spatial and temporal properties of the U.S. Latino population. Johansson and Bell (2014) looked at the economic aspect of music performance and identify the importance of a geographically rational manner in tour planning. As Hudson (2006) asserts, music reflects aspects of place and helps to shape place.

To summarize, a number of studies have articulated various aspects of music from a geographical perspective. Attention is usually drawn to the study of musical areas which cover a large geographic range comprising several distinct places. Unfortunately, most research did not further explore the implications of spaces and contexts for knowing musical activities. Duffy et al., (2016) pointed out the lack of and importance of geographic work in music studies. The reasons for geographers’ relative neglect of live music performances can be due to the data. For example, Mauch et al., (2015) identified that the major limiting factor in studying the evolution of popular music in the U.S. is the incomplete dataset. As a result, the exact geographic location of musical activities cannot be precisely delineated. With the increasing availability of data mining technologies, streaming music services can be a potential solution for overcoming the current data limitation (Morris & Powers, 2015).

2.2 Spatial analysis of music

During the last decade, emphasis has come to be placed on the importance of uncovering racial, social, economic, and historical production of music. Many researchers outlined their methodologies pointing to examinations of music’s role in creating a sense of place and its associated identities. Earlier publications have contributed to this topic by providing evidence for the active role of music scenes playing in the organization of social and economic space. Music scenes are generally interpreted as particular regions or cities with an agglomeration of musical activities (Peterson & Bennett, 2004). In Seman (2010)’s analysis, it was suggested to connect music scenes to a physical space in geographical studies. Music provides opportunities to gain new perspectives into various sectors of society. Whereas traditional musical geographical research is very much focused on a local art scene itself while less attention was paid to the context where a music scene formed.

More recently, it has been realized that music study needs to understand the emergence of a music scene. Studies have been put forward linking music events with local contexts that influence live music activities. For instance, budget, ticket price, and visitor capacity were found to be the key features of the music festival’s success in the Netherlands (Leenders et al., 2005). However, smaller case studies are not applicable to other places where influences may not be the same (Buck, 2016). Leenders et al., (2015) pointed to the importance of space when considering potentially event-driven changes. As widely recognized, relationships are not stationary across space and are said to exhibit spatial heterogeneity. Live music contributes in distinctive and dynamic ways to the characterization of cities (Gilstrap et al., 2021). However, spatial heterogeneity was seldom considered in music studies. As a result, in the music literature any phenomenon that is not stationary over space will not be represented well by a global model.

The awareness of the collective engagement of various industries with the music industry has been raised recently by academic studies of music. Bagley et al., (2022) recognized the importance of social and economic networks in a geographical study of music performance and production. The growing spatial value of live music in urban space is also evident in recent geographical discussions. What is useful about such an analysis is that it explores to what extent live music’s positive outcomes can contribute to the study of the relationships between people and built environments. Awareness of music localization can inform urban policy development on building interdisciplinary networks (Van der Hoeven & Hitters, 2020). For example, as sites of music scenes, cities benefit from music-based urban redevelopment projects (Seman, 2010). A lot of recent research has provided a localized perspective of music activities from a city level. However, the complexity of patterns of live music related to the subsector of economic development influenced by space was seldom explored.

The music industry is not homogeneous (Williamson & Cloonan, 2007). Watson et al., (2009) discussed the fact that the evolution of music was marked by distinct landscapes. Gibson (2002) suggested that the uneven geography of music is the reflection of a ‘city-country divide’ in demographic and economic profiles. It is important to note that live music is a critical resource for economic, social, and environmental developments in urban spaces (van der Hoeven & Hitters, 2023). This is complicated by the geographic context in relation to music. However, the existing literature failed to unveil the gap in the analysis of dynamics at the intersection of music and urban settings.

Traditionally the research community relied on limited survey data that reflected only a small group of musicians. This is mainly because it is hard to collect a massive volume of data about musician behavior (Oramas, 2016). With the ongoing development of Internet technologies, fostered by music streaming platforms country music was transformed into a major genre of American commercial music and its popularization was promoted to larger markets (Aguiar & Martens, 2016). The Application Programming Interface (API) service provides a programmatic gateway to download files stored in a database (Pereira et al., 2015; Yao & Wang, 2020). The concert information is increasingly made publicly available by music streaming platforms through API services and is available around the world (Eriksson et al., 2019). As such, researchers can rely on data sets collected from music streaming platforms of a magnitude richer than datasets used in previous music studies and, consequently, there will be a totally new way to look at musical activities in fine-grained detail (Li, 2022a).

From all of the above arises the need to research the spatial patterns of country music in urban areas. The theoretical and conceptual foundations of research on music events in space were laid by Florida et al., (2010) that pointed out that the geography of the music industry is dynamic, and a series of interacting factors contribute to this trend. In this regard, the main objective of this study is to identify which factors and to what extent these factors contribute to the distribution of country music events. In the review above of earlier findings regarding transformations in the geography of the music industry, few recreation and leisure studies have used regression models to detect the locally varying relationships between country music and its associated factors within urban areas. To address these methodological gaps in previous research, this study adopted the Geographically Weighted Regression (GWR) method to develop urban-level spatial regression models to offset the limitations of the aspatial regression method and suggest future methodological directions when using secondary data of music at the urban-level. The purpose of this paper therefore is to analyze the local geography of the relationship between country music events and the associated factors. This study used a two-step approach for regression modeling, in which the Ordinary Least Squares (OLS) regression was first applied and then the GWR was used. The remainder of the paper is structured as follows. The next section describes the data used in the analysis and elaborates briefly on the regression methods OLS and GWR. The results are presented in the fourth section. Implications of results are discussed in the fifth section. The last section concludes.

3 Data and methods

3.1 Data

The urban area was defined by the Census Bureau’s Urban–Rural Classification and comprises a densely developed territory that encompasses at least 2,500 people. Urban areas are categorized into two types: urbanized areas (UAs) with populations of 50,000 or more and urban clusters (UCs) with populations of at least 2,500, but fewer than 50,000. 3,054 UAs and 481 UCs appear to sparsely distribute across the contiguous U.S. (Fig. 1). Most UAs are concentrated along the coastlines and in major metropolitan areas. Cities such as New York City, Los Angeles, and Chicago comprise U.S.’s largest UAs and serve as economic and cultural hubs. Additionally, UAs serve as major intersections in transportation networks, including highways, airports, and train stations, facilitating connectivity and economic activities. In contrast, UCs, which are smaller in size with fewer populations, lie in suburban or rural regions, providing economic opportunities and transportation services for local communities. Overall, this complex distribution of UAs and UCs reflects the diverse socioeconomic and geographic landscapes that make up about 80% of the U.S. population (Ratcliffe et al., 2016).

Fig. 1
figure 1

(a) the number of country music concerts; (b) the number of fast-food outlets; (c) the number of facilities of air services; (d) the number of facilities of bus services; (e) the number of facilities of train services; (f) Top 10 cities with the most country music concerts (2009–2019) and distributions of urban areas (both UCs and UAs)

As one of the world's leading music streaming platforms, Spotify collects and manages an enormous amount of data. Spotify data highlights the concept of ‘big data’ due to its massive volume and diverse variety to allow deeper insights into music trends, artist popularity, and user experience. Over 100,000 country music events were chosen to be the sample for this study (Fig. 1). For convenience, Spotify Search API was utilized to extract concert data using pre-defined search queries according to several spatial constraints and keyword filters. The process involves several major steps. Firstly, the 1,000 most-streamed musicians on Spotify were retrieved from the musical artist database by using the tag 'country music'. Secondly, the concerts where those musicians performed between 2009 and 2019 were located. The focus was then placed on concerts that only took place within the contiguous U.S. Generally, country music concerts are prevalent in the Southern states. Cities such as Nashville, Dallas, New York City, and Austin, known for their music scenes, attract a significant number of music events. It was observed that over 85% of the concerts occurred in urban areas. UAs tend to have a higher concentration of concerts than UCs. For those concerts not in urban areas, it was found that a majority of them were hosted in venues such as fairgrounds, resorts, state parks, and casinos, which offered a wide range of leisure services beyond hosting music events. Therefore, non-urban concerts were excluded from the dataset, allowing for a more targeted analysis of country music concerts only in urban areas.

Southern food and country music are integral components of the rich and vibrant culture found in the South. Both reflect the region's history, traditions, and diverse influences (Smith, 1980). Considering that both of them as two cultural expressions of the South are potentially intricately intertwined, this study aims to gain insights into catering facilities serving Southern food and their connectivity to concert locations by using the Point of Interest (POI) dataset. Over 16,000 addresses were geocoded to map the spatial distribution of Southern fast-food restaurants. To ensure chain diversity, 18 different Southern fast-food chains were selected, including Bojangles, Checkers & Rally's, Chick-Fil-A, Church's Chicken, Cook Out, Cracker Barrel, Hardee's, Jack's Fast Food, Jason's Deli, Krystal, McAlister's Deli, Newk's Eatery, Raising Cane's Chicken Fingers, Sonic, Waffle House, Whataburger, Wingstop, and Zaxby's. The distribution of fast-food outlets with Southern roots reflects that they have a significant presence across the whole U.S. and dominate major cities especially those of Southern states such as Texas and Georgia (Fig. 1).

Assessing the interplay between transportation accessibility and music events is helpful in identifying areas with robust transportation links that may attract larger audiences (Chang et al., 2022). The transportation facility data was obtained online from the U.S. Department of Transportation. The dataset contains facilities of bus services, train services, and air services as of 2018 separately. Specifically, facilities of bus services are intercity bus stops with scheduled service (excluding charter bus and commuter bus services), which are from the Bureau of Transportation Statistics' (BTS') Intermodal Passenger Connectivity Database (IPCD); facilities of train services are Amtrak and Alaska rail facilities with scheduled service, which are also from the BTS’ IPCD; facilities of air services refer to commercial airports defined by the Federal Aviation Administration (FAA).

The number of country music concerts (C), the number of fast-food chains (res), and the number of facilities of air services (air), bus services (bus), and train services (rail) were calculated for each urban area. A description of the variables is shown in Table 1. All five variables have large proportions in UAs compared to UCs. Specifically, over 90% of country music concerts, more than 80% of fast-food outlets, and at least 64% of transportation service facilities are concentrated in UAs (Fig. 2). Using total counts instead of normalizing total counts by local population can facilitate meaningful comparisons across different urban areas, which avoids underrepresenting areas with smaller populations that may still host music events. The number of country music events was then regressed on the number of Southern fast-food chains and the number of transportation service facilities.

Table 1 Description of the data
Fig. 2
figure 2

(a) the proportion of country music concerts; (b) the proportion of fast-food outlets; (c) the proportion of facilities of air services; (d) the proportion of facilities of bus services; (e) the proportion of facilities of train services

3.2 Methods

The paper presents a spatial analysis of patterns of music events, based on the mapping of country music concerts and the interpretation of the obtained modeling results. It included two phases. In the first phase of regression analysis, OLS was conducted to regress the number of concerts against the number of fast-food outlets and the number of transportation facilities in each urban area. In the second phase, the spatial distribution of concerts was analyzed in detail—the location of concerts was considered, and then their relationships with fast-food outlets and transportation facilities were analyzed by using GWR. For the purpose of comparability of the model performance of different regression methods, R2, adjusted R2, and AICc measures were used.

The general form of an OLS model can be presented as (Ward & Gleditsch, 2018):

$$\mathrm{y}=\alpha +{\sum }_{j=1}^{n}{x}_{j}{\beta }_{j}+\varepsilon$$

In Eq. (1), α denotes the intercept, \({\beta }_{j}\) denotes the regression coefficient for the jth explanatory variable, n denotes the number of explanatory variables, and \(\varepsilon\) denotes a random error. Assumptions of the OLS regression are that observations are independently distributed, and errors are normally distributed and heteroscedastic. If these assumptions hold true, the regression coefficient can be estimated by (Brunsdon et al., 1996):

$$\widehat{\beta }={\left({X}^{T}X\right)}^{-1}{X}^{T}y$$

The GWR model has the capability of dealing with spatial nonstationarity by assuming that the spatial phenomenon varies across a geographical area. For each location i, a standard GWR model can be denoted by (Brunsdon et al., 1996):

$${\mathrm{y}}_{i}=\alpha ({u}_{i},{v}_{i})+{\sum }_{j=1}^{n}{x}_{ij}{\beta }_{j}({u}_{i},{v}_{i})+{\varepsilon }_{i}$$

In Eq. (3), ui and vi are spatial coordinates at the ith location, α denotes the local intercept, \({\beta }_{j}\) denotes the regression coefficient at the ith location, and εi denotes the random error at the ith location and is assumed to be independently and identically distributed. The regression coefficient for each explanatory variable at the ith location is estimated from:

$${\widehat{\beta }}_{i}\left({u}_{i},{v}_{i}\right)={({X}^{T}{W}_{i}\left({u}_{i},{v}_{i}\right)X)}^{-1}{X}^{T}{W}_{i}\left({u}_{i},{v}_{i}\right){y}_{i}$$
$${W}_{i}=\left[\begin{array}{ccc}{w}_{i1}& \cdots & 0\\ \vdots & \ddots & \vdots \\ 0& \cdots & {w}_{im}\end{array}\right]$$

In Eq. (4) and (5), \({\widehat{\beta }}_{i}\) is the estimate of the regression coefficient for the jth explanatory variable, X is the matrix of explanatory variables, m is the number of surrounding locations of location i, Wi is a distance decay spatial weights matrix whose elements wi1, …, wim were calculated by a weighting function (Fotheringham et al., 2002):


In Eq. (6), wij describes the level of interdependence between location i and location j. An adaptive Gaussian kernel function is specified with a distance (dij) between location i and location j and a bandwidth (b) determined using cross-validation.

Once all models have been implemented, the Local Indicators of Spatial Autocorrelation (LISA) analysis will be utilized to examine spatial patterns of residuals of both OLS and GWR models. LISA analysis is able to identify spatial patterns where the residuals exhibit similarity in their values significantly or not, indicating the presence or absence of spatial autocorrelation (Anselin, 2020). This study considers five types of spatial patterns: high values neighboring to other high values (High–High), high values neighboring to low values (High–Low), low values neighboring to high values (Low–High), low values neighboring to other low values (Low–Low), and no indication of significant local spatial clustering (not significant). The LISA analysis results can help assess if a GWR model has successfully accounted for spatially structured variations that may remain unexplained by an OLS model and effectively reduced spatial autocorrelation in the dataset.

4 Result

All OLS models were first tested to determine the overall fit as well as influence of each variable on the dependent variable. To test multicollinearity between independent variables, the Variance Inflation Factor (VIF) among the candidate predictors was calculated. The highest VIF observed was below the danger level of 7.5. Thus, there is no multicollinearity. Table 2 presents the summary of regression statistics. In general, the OLS method effectively explained 61% of variations in model residuals. Explanatory variables are statistically significant with the anticipated positive signs. More specifically, each variable exhibits a distinct association with the number of country music concerts. An increase in the number of fast-food chains relates to an increase in the number of country music concerts, implying that the city’s food service industry contributes to its reputation as an attractive destination for music events. Among the three dimensions of transportation services, the air service demonstrates the highest regression coefficient with the number of country music concerts. This is most likely due to fewer air service facilities compared to bus stations or train stations. As indicated by a regression coefficient of 49.7, the number of facilities of bus services positively relates to the number of country music concerts. Similarly, the number of facilities of train services is also positively related to the number of country music concerts with a regression coefficient of 160.19. The relative strengths of the regression coefficients obtained from univariate regression models (i.e., OLS RES, OLS AIR, OLS RAIL, and OLS BUS) remained consistent in the multivariate model (i.e., OLS FULL). Specifically, in the OLS FULL model, the number of fast-food chains associates with the smallest coefficient (4.10), the air service exhibits the highest coefficient (42.08), and the train service has a coefficient (34.15) larger than the bus service (21.11). This observation suggests that each explanatory variable has a robust relationship with the dependent variable. Moreover, the consistency in signs of coefficients indicates that the positive relationship remains stable despite variations in their magnitudes across univariate and multivariate models.

Table 2 Results of regression analyses

To clearly understand model performance for different transportation types, the R2 and AICc of three different transportation facilities were compared. In OLS results, the bus facilities have been shown to contribute more greatly to the modeling of event variations than air or train facilities. The outcomes of the GWR models reveal improvements over the OLS models. The R2 (0.8) and adjusted R2 (0.79) for the GWR RES model are higher than R2 (0.5) and adjusted R2 (0.5) for the OLS RES model while AICc (41,642.46) for the GWR RES model is lower than AICc (44,663.90) for the OLS RES model. The R2 and adjusted R2 for the OLS AIR model, OLS BUS model, and OLS RAIL model were 0.26 and 0.26, 0.3 and 0.3, 0.23 and 0.23 whereas the GWR AIR model, GWR BUS model and GWR RAIL model produced R2 of 0.38 and adjusted R2 of 0.37, R2 of 0.46 and adjusted R2 of 0.45, R2 of 0.27 and adjusted R2 of 0.27, separately. For the OLS FULL model, the R2, adjusted R2, and AICc were 0.61, 0.61, and 43,858.50, while for the GWR FULL model the R2, adjusted R2, and AICc were improved to 0.87, 0.86, and 40,213.74. These improvements suggest that the GWR model demonstrates a significantly better explanation of the data, after taking into account the data’s spatial patterns that the OLS model overlooked. An examination of Fig. 3 shows that the local R2 illustrates a considerable spatial variation of concert events explained by the model. As shown in Fig. 3, the local R2 of the RES model for most urban areas is estimated at a higher value compared to the R2 estimated by the AIR model, BUS model, or RAIL model. Some Southeast states, for example, Georgia, South Carolina, and South Carolina show up as a region with R2 values larger than 0.75 for the RES model compared to the AIR model, BUS model, and RAIL model showing R2 values less than 0.75. In addition to this, the Northeast exhibits higher R2 values for the BUS model than for the AIR model or RAIL model. The West, Midwest, Northeast, and South all have higher local R2 values for the FULL model compared to the RES model, AIR model, BUS model, or RAIL model.

Fig. 3
figure 3

Local R-squared (R2) values for (a) RES model; (b) AIR model; (c) BUS model; (d) RAIL model; (e) FULL model

Figure 4 shows that coefficients of explanatory variables vary significantly across the geographical area, which is hidden in the OLS models. As can be seen on the map, the regression coefficients of transportation facilities are estimated to be between 46.61 and 307.92 for air services, 36.07 and 88.155 for bus services, and 69.36 and 172.15 for train services in the West Coast urban areas, whilst there are some urban areas located in Texas where regression coefficient is estimated at a much higher value between 516.81 and 855.41 for air services, 179.91 and 263.39 for bus services and 307.58 and 4475.88 for train services. The relationship between country music events and fast-food chains also varies geographically, that is, in terms of magnitude, the number of country music events in Northwest urban areas has strong ties with the number of fast-food chains and in other regions such as the South has weak ties with the number of fast-food chains.

Fig. 4
figure 4

Regression coefficients for (a) RES model; (b) AIR model; (c) BUS model; (d) RAIL model

The findings above lead to the conclusion that Southern fast-food outlets play greater roles in explaining the distribution of country music events as compared to the transportation service facilities. A comparative analysis of local R2 values (Fig. 5) shows that the RES model has equivalent model performance in both UAs and UCs; while the AIR model, BUS model, and RAIL model exhibit better model performance in UAs than in UCs. Additionally, the FULL model, which represents the best fit among the models, yields similar model performances in UAs and UCs. This finding indicates that the FULL model captures the underlying variations of music events across two types of urban areas effectively, implying the importance of considering multiple factors in comprehensive models.

Fig. 5
figure 5

Boxplots of local R2 values for (a) RES model; (b) AIR model; (c) BUS model; (d) RAIL model; (e) FULL model

LISA analysis of patterns of model residuals further enhances understanding of model improvements achieved through the implementation of the GWR method. Figure 6a and b present the distribution of significant and insignificant spatial autocorrelation of residuals of the OLS FULL model and GWR FULL model, respectively. For the OLS FULL model, Texas and Midwestern states exhibit several High-High clusters where high model residuals are surrounded by high model residuals, indicating positive spatial autocorrelation. Low-Low clusters of the OLS FULL model’s residuals are present in several Southeastern states. For GWR FULL model, an isolated High-High cluster of residuals is observed in San Antonio located in south-central Texas. High-Low and Low–High patterns of the GWR FULL model’s residuals are less pronounced compared to the ones of the OLS FULL model’s residuals. Overall, the GWR method significantly reduced model residuals generated by the OLS method.

Fig. 6
figure 6

(a) LISA statistics map of OLS FULL model’s residuals; (b) LISA statistics map of GWR FULL model’s residuals

5 Discussion

This study highlights substantial effects of locational attributes on music agglomeration that were observed in urbanized areas frequently (Nunes & Birdsall, 2021). This is because the urban areas are very well-established areas with a good number of amenities as compared to the rural areas (Herrero et al., 2006). Artistic and cultural producers are densely agglomerated in urbanized areas, which are acknowledged to be more productive in music production (London, 2017). Historically, Los Angeles and New York City have been considered the major music centers of live entertainment, in that they are large cities offering large markets for music employment and consumption (Baker, 2019; Florida et al., 2010). After a period of structural changes in the global economy, some mid-sized cities have taken center stage as livable sites of culture and new economy jobs (Florida & Jackson, 2010). This can explain why Nashville is the center of country music performance, broadcasting, and recording (Pecknold, 2014). The above finding is in line with the study carried out by Baker (2016) in the United Nations Educational, Scientific and Cultural Organization’s (UNESCO’s) Creative Cities Network where they found musical activities are shaped by the geographic restructuring of the music industry. Spaces of entertainment were being developed differently in city centers and on the outskirts (Wynn, 2015). The concentration of facilities is primarily centered in the city's urban area, with a diminishing density as one moves toward the periphery (Krasulski & Harlow, 2010). For example, Nashville downtown area hosts a concentration of facilities for local music businesses (Johansson, 2010). In addition to the inner city, locations such as revitalized areas of former industrial enterprises in outer metropolitan suburbs are also home to music festivals (Rosentraub, 2014). Figure 4 implies that changes in the spatial structure of regression coefficients have included the commercialization of urbanized areas of cities such as Nashville and Dallas, while gentrification and the emergence of new hotspots of musical activities are also found in urban clusters of some cities such as Nacogdoches and Branson. This finding implied decentralized forms of relationships between music events and food and transportation service facilities, although places such as Nashville were known for their prominent roles in the country music industry where music performances and productions agglomerated.

This study's findings unveil that urban areas of cities such as Nashville, Dallas, and Austin are home to concerts that are with relatively higher correlations with the number of transportation facilities. The main reason is that public transportation is an important preference considered in the choice of concerts to attend (Steinfeld & Steinfeld, 2017). Ross (2017) provides evidence that live music's contribution to the urban area is facilitated by the availability of public transportation. Cities’ centrality in the network aids urban employment growth through well-developed transportation infrastructures (Neal, 2011). A well-connected transportation network within the urban setting generates economic growth and is appealing to major corporations, firms, and entrepreneurs for locating there. The significance of transportation is particularly evident in densely populated urban areas with high traffic volumes, where a wide array of amenities and spaces were designed to accommodate various consumer activities such as shopping and recreation (Sallis et al., 2016). In this regard, big events with massive crowds of people received special attention in transportation planning practices in some large cities (Pereira et al., 2015). In areas with high population density, the number of transportation facilities should be large accordingly in order to meet the high demand for commuters. As a result, the proximity to transportation facilities intensifies social relations and provides unique advantages for accessing live music in urbanized areas (Whiting & Carter, 2016).

In this study, the number of fast-food outlets is significantly related to the country music concert distributions. This finding implies that music events have the potential to generate spillover effects on economic development by optimizing local economic benefits from the catering industry development. The food sector plays a vital role in shaping the economic and social fabric of a region and is a particularly important sector for employment in many cities (Omholt, 2015). The ‘music city’ is the result of music industry growth that emerged from investment in the accommodation and food services industry (Beckman et al., 2013). Nashville as an exemplary music city, provides a supportive environment for the promotion and development of country music (Pecknold, 2014). Followed by the explosion of catering, restaurants are attractive spaces for a great time in Nashville (Benzmiller, 2008). Restaurants are often located near consumption spaces. For example, urban areas are characterized by the concentration of shopping areas such as attractive streets and shopping malls that provide catering services (Johansson, 2010). The cause of this pattern lies partly in the needs of a large number of consumers (Hedhli et al., 2013). The catering demands of concert attendance have led to the emergence of restaurants and other food service establishments (Scott & Harmon, 2016). In accordance with these findings above, this study implies that the pattern of the spatial distribution of restaurants reflects food culture and overall lifestyle and well-being of the residents of music cities.

This study finds that country music not only maintained a popular presence in Southern cities such as Nashville, Dallas, and Austin but also in traditional music markets such as New York City, Los Angeles, and Chicago (Li, 2022a). Scholars have traditionally associated country music with the American South (Meier, 2019), while also observing regional variations in its popularity gradually (White & Day, 1997). This can be explained by the urbanization in American cities (Fay, 2014). Country music had its beginnings in the South and has blossomed in many areas (Gibson, 2014). As a commercial product of modern industrial America, country music is no longer targeting a specific consumer market culturally or geographically (Ziewacz, 2001). The unique geographic characteristics of cities have led to a series of profound social, economic, cultural, and demographic changes, while the industrial production structure is diversified (Oakes & Warnaby, 2011). Benefiting from their initial foundations as the center of service-driven economies, some cities were undergoing socioeconomic and cultural changes that spread across different industries and sectors such as the live music industry (Benzmiller, 2008). Some cities are competing for attracting music artists and retain investments to market themselves as tourist destinations (Long, 2014). The main reason is that those attending the concerts are more likely to be tourists (Lashua et al., 2014). Cities embrace larger and more diversified music markets as they can benefit from growing and diversifying tourist masses (Brandellero & Janssen, 2014).

This study utilized an approach combining univariate and multivariate regression modeling, which allows for a rigorous investigation of the relationships between the dependent variable and explanatory variables, offering a more nuanced understanding of individual and collective contributions of explanatory variables to the explained variability in the model. Variables for both univariate and multivariate regression models were selected on the basis of their inclusion in previous studies. Expenses for food and transport are two main parts of concert attendance spent at the event (Behr et al., 2020a, 2020b). However, they were seldom discussed in a spatial setting. This study confirmed the initial assumption that the catering and transportation of the neighborhood are important spatial determinants of entertainment activities. Music festivals tend to be spatially dependent (Van Aalst & Van Melik, 2012). This information is completely invisible in global estimation and is only seen in the local model. Previous research has demonstrated that spatial nonstationarity is an important consideration when investigating the evolution of change (Li, 2022b). The results depicted in Table 2 and in Fig. 1 confirm that spatial interaction is a relevant and important aspect of change and underscore the importance of using GWR analysis in the study of music geography. OLS results illustrate the generality of how the number of country music events is related to the number of food and transportation service facilities. The OLS model was extended by the GWR model incorporating spatial effects that were modeled as spatial heterogeneity. There is a marked improvement of using GWR over OLS results highlighted by both model performance indicators and LISA analysis results. The GWR model behaves more efficiently and enables in-depth local analysis to be conducted because it uses a locally adjusted regression equation for each observation (Li, 2022a; Nakaya, 2001). Considering spatial nonstationarity provides greater insight into distributions of country music events and their relations to food service and transportation facility locations.

A large part of the existing music geography literature makes use of aggregate data such as interviews and reports but was limited to access to data with fine spatial resolution. This study fills this void by using a ‘big data’ source: GPS-tagged concert information retrieved from Spotify. Leveraging such high-location-accuracy data enables the exploration of disparities in musical activities between urbanized areas and urban clusters. The study thus expands the analytical scope beyond previous works. The overarching outcome of the presented research is that further study does not need expensive proprietary data to estimate and predict musical behaviors, instead can integrate the ‘big data’ fetched through API services to obtain more fine-grained results (e.g., at an individual level). The increase in the availability of API-powered ‘big data’ is clearly important for the future of spatial music studies.

This study highlights that because of food service density and close proximity to public transportation facilities, there is a concentration of resources (e.g., catering, accommodation, sponsors, and tourists) which builds competitive advantages for cities to host country music events. In terms of practical applications, a notable implication arising from findings regarding concert distributions reported here is that urban planning needs to sustain accessible food and public transportation services in the community so that music events can be viable investments in the transition of a city to a ‘music city’. In this respect, investigations on country music in urban areas are likely to be of broad interest to scholars concerned about the role of music in urban development. This study has obtained a general picture of the distribution of music events in the U.S. and identified food and transportation sectors that are related to the spatial patterns of music events in the pre-COVID-19 period. The entertainment landscape witnessed a revival in 2022 including the resumption of live music, it is important to recognize that uncertainties in the urban industrial ecosystem may still persist during the post-COVID-19 era. Nonetheless, the conclusions drawn from this study are believed to remain relevant and have implications for future research conducted in similar contexts.

6 Conclusions

Food, transportation, and the entertainment industry are all indelible parts of the fabric of the urban setting. They underpin cities such as Nashville, Dallas, New York City, and Austin for being leading host sites of country music events. This study presented explanatory capabilities of food and transportation service dimensions to infer country music events at the urban level. It was demonstrated that the aggregation of food services and dense transportation facilities lead to clusters of country music events in urban areas. The cases of Nashville, Dallas, and Austin illustrate how cities, by being the food and transportation hubs in the South have become the country music city that leads in the sites nationwide where numbers of country music events were held.

The general outcome of the comparative study of the two regression methods is that GWR is superior to OLS. In terms of its overall findings. The results of both OLS and GWR demonstrated that both food and transportation services played a significant role in explaining the distributions of country music events. With the presence of definitive regional patterns of country music events in relation to Southern fast-food chains and transportation facilities, the findings suggest regional developers and policymakers utilize the knowledge gained in this research to improve the city’s competitiveness and increased its reputation after engaging in “music-led” strategic development.