Major American Sports Venues

Abstract

The four major North American sports leagues are embedded throughout the cultures of the United States and Canada. Sports venues such as Wrigley and Fenway are often as recognizable of landmarks as any within their respective cities. The purpose of this analysis is to explore the venues associated with the four major North American sports, with the ultimate goal of creating an interactive leaflet map that allows a user to click and explore the venues for themselves.

A rendered copy of this RMarkdown document can be found at my rpubs account.

# loads revelant libraries
library(rvest)
library(tidyverse)
library(knitr)
library(kableExtra)
library(leaflet)

Getting and Cleaning Data

Wikipedia conveniently has pages containing venue information for the four major North American sports.

# We will pull our venue data from four different wikipedia pages
urls = c('https://en.wikipedia.org/wiki/List_of_current_Major_League_Baseball_stadiums',
         'https://en.wikipedia.org/wiki/List_of_National_Basketball_Association_arenas',
         'https://en.wikipedia.org/wiki/List_of_current_National_Football_League_stadiums',
         'https://en.wikipedia.org/wiki/List_of_National_Hockey_League_arenas')

We can use the rvest package to scrape the relevant venue information from each page, manually associating each dataset with the appropriate sports league as we go.

# MLB Data

mlbVenues = read_html(urls[1])

mlbVenues = mlbVenues %>%
    html_nodes('table') %>%
    .[[2]] %>%
    html_table() %>%
    as_tibble()

mlbVenues = mlbVenues %>%
    select(Name, Capacity, Location, Team, Opened) %>%
    mutate(League = 'MLB')

# NBA Data

nbaVenues = read_html(urls[2])

nbaVenues = nbaVenues %>%
    html_nodes('table') %>%
    .[[1]] %>%
    html_table() %>%
    as_tibble()

nbaVenues = nbaVenues %>%
    select(Arena, Capacity, Location, `Team(s)`, Opened) %>%
    mutate(League = 'NBA') %>%
    rename(Team = `Team(s)`, Name = Arena)

# NFL Data

nflVenues = read_html(urls[3])

nflVenues = nflVenues %>%
    html_nodes('table') %>%
    .[[2]] %>%
    html_table() %>%
    as_tibble()

nflVenues = nflVenues %>%
    select(Name, Capacity, Location, `Team(s)`, Opened) %>%
    mutate(League = 'NFL') %>%
    rename(Team = `Team(s)`)

# NHL Data

nhlVenues = read_html(urls[4])

nhlVenues = nhlVenues %>%
    html_nodes('table') %>%
    .[[1]] %>%
    html_table() %>%
    as_tibble()

nhlVenues = nhlVenues %>%
    select(Arena, Capacity, Location, `Team(s)`, Opened) %>%
    mutate(League = 'NHL') %>%
    rename(Team = `Team(s)`, Name = Arena)

Now that we’ve extracted each table into its own separate dataframe, we’ll combine into a common dataset.

# Binds the information into a single tibble
venues = rbind(mlbVenues, nbaVenues, nflVenues, nhlVenues)
venues %>%
    head(10) %>%
    kable() %>%
    kable_styling(bootstrap_options = "striped", full_width = F)

Name	Capacity	Location	Team	Opened	League
Angel Stadium	45,517[1]	Anaheim, California	Los Angeles Angels	1966	MLB
Busch Stadium	45,494[2]	St. Louis, Missouri	St. Louis Cardinals	2006	MLB
Chase Field	48,686[3]	Phoenix, Arizona	Arizona Diamondbacks	1998	MLB
Citi Field	41,922[4]	Queens, New York	New York Mets	2009	MLB
Citizens Bank Park	42,792[5]	Philadelphia, Pennsylvania	Philadelphia Phillies	2004	MLB
Comerica Park	41,083[6]	Detroit, Michigan	Detroit Tigers	2000	MLB
Coors Field	46,897[7]	Denver, Colorado	Colorado Rockies	1995	MLB
Dodger Stadium	56,000[8]	Los Angeles, California	Los Angeles Dodgers[nb 2]	1962	MLB
Fenway Park	37,755[9]	Boston, Massachusetts	Boston Red Sox[nb 3]	1912	MLB
Globe Life Park in Arlington	48,114[10]	Arlington, Texas	Texas Rangers	1994	MLB

There are some formatting issues that we’ll have to address:

The Location variable is currently a single text field that contains both city and state. We’d like to extract city and state into separate variables.
The Name, Team, Capacity, and Opened variables have some stray brackets ([.*]) embedded within the values for certain records. We’ll remove any bracketed data throughout these fields.
Once brackets have been removed, the Capacity and Opened variables will need to be converted to integers.

# Creates city and state features and coerces data into correct types
venues = venues %>%
    separate(Location, c('City', 'State'), ', ') %>%
    mutate(Name = str_replace(Name, '\\[.*\\]', ''),
           Team = str_replace(Team, '\\[.*\\]', ''),
           Capacity = as.integer(str_replace_all(str_replace(Capacity, '\\[.*\\]', ''), ',', '')),
           Opened = as.integer(str_sub(Opened, 1, 4)))

Now that we’ve addressed our basic formatting issues, we’ll perform a sanity check to make sure we’re not missing any venues. Each league currently has the following number of teams:

30 MLB Teams
30 NBA Teams
32 NFL Teams
31 NHL Teams

# Summarizes the venue counts by League for sanity checks
venues %>% 
    group_by(League) %>%
    summarise(VenueCount = n()) %>%
    kable() %>%
    kable_styling(bootstrap_options = "striped", full_width = F)

League	VenueCount
MLB	30
NBA	30
NFL	31
NHL	32

Our team and venue counts for the MLB and the NBA match exactly, but it looks like we have one less venue for the NFL and one more venue for the NHL. Finding the extra NFL record shows us that the two New York teams share a football stadium. We’ll convert these into separate records.

# Two nfl teams share the same stadium
venues %>% 
    filter(League == 'NFL') %>%
    mutate(TeamLength = sapply(Team, str_length)) %>%
    arrange(desc(TeamLength)) %>%
    top_n(1) %>%
    select(-TeamLength) %>%
    kable() %>%
    kable_styling(bootstrap_options = "striped", full_width = F)

Name	Capacity	City	State	Team	Opened	League
MetLife Stadium	82500	East Rutherford	New Jersey	New York GiantsNew York Jets	2010	NFL

# Creates separate rows for each team
venues = venues %>%
    mutate(Team = ifelse(Name == 'MetLife Stadium', 
                         'New York Giants', 
                         Team))
venues = rbind(venues,
               venues %>%
                   filter(Name == 'MetLife Stadium') %>%
                   mutate(Team = 'New York Jets'))

In the NHL, the New York Islanders have two active venues. We’ll note this fact but will leave our data as is.

# One NHL team has two stadiums
venues %>%
    group_by(Team) %>%
    summarise(NumberOfVenues = n()) %>%
    filter(NumberOfVenues > 1) %>%
    inner_join(venues, by = c('Team' = 'Team')) %>%
    select(-NumberOfVenues) %>%
    kable() %>%
    kable_styling(bootstrap_options = "striped", full_width = F)

Team	Name	Capacity	City	State	Opened	League
New York Islanders	Barclays Center	15795	Brooklyn	New York	2012	NHL
New York Islanders	Nassau Coliseum	13900	Uniondale	New York	1972	NHL

Exploratory Analysis

Next we’ll perform some descriptive analysis on our dataset. We’ll start with some basic statistics about venue capacity by league.

venues %>%
    group_by(League) %>%
    summarise(MeanCapacity = mean(Capacity),
              SdCapacity = sd(Capacity),
              MedianCapacity = median(Capacity),
              MaxCapacity = max(Capacity),
              MinCapacity = min(Capacity)) %>%
              kable() %>%
              kable_styling(bootstrap_options = "striped", full_width = F)

League	MeanCapacity	SdCapacity	MedianCapacity	MaxCapacity	MinCapacity
MLB	42517.33	5598.775	41907.5	56000	25000
NBA	18930.57	1049.816	18987.5	20917	16867
NFL	69472.19	10238.696	69137.5	82500	27000
NHL	18082.38	1411.292	18288.5	21302	13900

ggplot(venues, aes(League, Capacity, fill = League)) +
    geom_boxplot() +
    labs(title = 'Venue Capacity by League',
         x = 'League',
         y = 'Capacity') +
    theme(plot.title = element_text(hjust = .5))

ggplot(venues, aes(x = Capacity, fill = League)) +
    geom_histogram() +
    labs(title = 'Venue Capacity by League',
     x = 'Capacity',
     y = 'Count') +
    facet_wrap(~League, ncol = 2, scales = 'free_x') +
    theme(plot.title = element_text(hjust = .5))

NFL teams clearly tend to have the largest stadium sizes, followed by the MLB teams. NBA teams seem to only narrowly have a greater capacity than NHL teams. This shouldn’t be shocking. About a 3rd of NHL and NBA teams share venues, so aside from some small seating re-arrangements, the two leagues should have similar capacity characteristics.

sharedVenues = venues %>%
    group_by(Name) %>%
    summarise(TeamCount = n()) %>%
    filter(TeamCount > 1) %>%
    inner_join(venues, by = c('Name' = 'Name'))

ggplot(sharedVenues, aes(League, Name, col = League)) +
    geom_point(pch = 'x', size = 8) + 
    labs(title = 'Venues Shared by Multiple Teams',
         x = 'League',
         y = 'Venue') +
    theme(plot.title = element_text(hjust = .5))

Next we’ll shift our focus over to the what year the stadiums were opened.

venues %>%
    mutate(outlierCaption = ifelse(Opened < 1965, paste0(Name, ', ', Opened), '')) %>%
    ggplot(., aes(League, Opened, fill = League)) +
    geom_boxplot() + 
    geom_text(aes(label = outlierCaption), hjust = -.2, size = 2) + 
    labs(title = 'Year Opened by League',
         x = 'League',
         y = 'Year Opened') +
    scale_y_continuous(breaks = seq(1910,2020,10)) + 
    theme(plot.title = element_text(hjust = .5))

The MLB has the two oldest stadiums in Fenway Park and Wrigley Field, followed by the Los Angeles Memorial Coliseum and Soldier Field in the NFL.

It would be interesting to see if the capacity of a stadium has anything to do with the year in which it was built. We explore this possibility in the below plots.

ggplot(venues, aes(Opened, Capacity, col = League)) +
    geom_point(alpha = .7) +
    geom_smooth(method = 'lm', lty = 2, size = .7) +
    facet_wrap(~League, ncol = 2, scales = 'free')

There’s not a clear discernable relationship between the year that a venue opened and the seating capacity. Our regression lines for the MLB, NFL, and NHL are near flat. Our NBA regression line trends downward, but the confidence bands include the possibility of a slope of zero.

Stadium Locations of Major American Sports Leagues

Next we’d like to visualize the locations of the different sport venues. Our existing dataset contains information about city and state, but we can be more precise by pulling in latitudinal and longitudinal values.

# Reads csv of coordinate data
coordPath = 'C:/Users/Rick/Documents/Projects/Sports_Venues/venueCoordinates.csv'
coordinates = read.csv(coordPath, stringsAsFactors = F)
head(coordinates) %>% 
    kable() %>%
    kable_styling(bootstrap_options = "striped", full_width = F)

sport	lat	lon	team_name	channel
NBA	25.7814	-80.1878	Miami Heat	Fox Sports Sun
NBA	28.5392	-81.3836	Orlando Magic	Fox Sports Florida
NBA	29.4269	-98.4375	San Antonio Spurs	Fox Sports Southwest
NBA	29.7508	-95.3622	Houston Rockets	Root Sports Southwest
NBA	29.9489	-90.0822	New Orleans Pelicans	Fox Sports New Orleans
NBA	32.7904	-96.8103	Dallas Mavericks	Fox Sports Southwest

Joining our coordinate data with our venue data allows us to generate an interactive leaflet map of the stadium locations, color coordinated by league.

venues = venues %>% 
    left_join(coordinates, by = c('Team' = 'team_name')) %>%
    rename(Latitude = lat, Longitude = lon) %>%
    select(League, Name, Capacity, City, State, Team, Opened, Latitude, Longitude)

pal <- colorFactor(c("red", "darkgreen", "darkorange", "blue"), levels(venues$League))
popHTML = paste0('<dl>',
                 '<dd>Venue: ', venues$Name, '</dd>',
                 '<dd>Location: ', venues$City, ', ', venues$State, '</dd>',
                 '<dd>Team: ', venues$Team, '</dd>',
                 '<dd>Capacity: ',  venues$Capacity, '</dd>',
                 '</dl>')
venues %>% 
        leaflet() %>% 
        addTiles() %>% 
        addCircleMarkers(clusterOptions = markerClusterOptions(), 
                popup = popHTML, 
                color = ~pal(venues$League), 
                fillOpacity = 1,
                radius = venues$Capacity/6000) %>% 
        addLegend(
                pal = pal,
                values = venues$League)

Major American Sports Venues

Erik J. White

Abstract

Getting and Cleaning Data

Exploratory Analysis

Stadium Locations of Major American Sports Leagues