Global Coffee Chains

Fueled By Dunkin

Week 6 - Global coffee-chain locations (as of 2017 or 2018)

RAW DATA
Article
DataSource (Starbucks): kaggle.com
DataSource (Tim Horton): timhortons.com
DataSource (Dunkin Donuts): odditysoftware.com

require(readxl)
require(dplyr)
require(maps)
require(ggmap)
require(fields)
require(sf)

setwd("~/Desktop/tidytuesday/data/2018/2018-05-07")

coffee1 <- read_excel("week6_coffee_chains.xlsx", sheet = 1)
coffee2 <- read_excel("week6_coffee_chains.xlsx", sheet = 2)
coffee3 <- read_excel("week6_coffee_chains.xlsx", sheet = 3)

## all have different columns

starbucksUS <- subset(coffee1, Country == "US")
dunkinUS <- subset(coffee3, e_country == "USA") ## all in US so not really needed

How far do I need to walk/drive to get my fix?

I’m putting my personal biases into this, so I"m only looking at Dunkin and Starbucks within the US.

getMeToCoffee <- function(myPoint, modePrefer, coffeePrefer) {
  if (is.character(myPoint)) {
    myPoint2 <- geocode(myPoint, output = "latlon")
  }

  if (coffeePrefer == "dunkin") {
    toDunkin <- rdist(as.matrix(myPoint2), dunkinUS[, c("loc_LONG_centroid", "loc_LAT_centroid")])
    closestDunkin <- which.min(toDunkin)
    tryThis <- paste(dunkinUS$"e_address"[closestDunkin], dunkinUS$e_city[closestDunkin], ",", dunkinUS$e_state[closestDunkin], sep = " ")
    test <- mapdist(myPoint, tryThis,
      mode = modePrefer,
      output = "simple", messaging = FALSE, sensor = FALSE,
      language = "en-EN", override_limit = FALSE
    )
    ## does better with character addresses than lat longs
  } else if (coffeePrefer == "starbucks") {
    toStarbucks <- rdist(myPoint2, starbucksUS[, c("Longitude", "Latitude")])
    closestStarbucks <- which.min(toStarbucks)

    tryThis <- paste(starbucksUS[closestStarbucks, "Street Address"], starbucksUS$City[closestStarbucks], ",", starbucksUS[closestStarbucks, "State/Province"], sep = " ")
    test <- mapdist(myPoint, tryThis,
      mode = modePrefer,
      output = "simple", messaging = FALSE, sensor = FALSE,
      language = "en-EN", override_limit = FALSE
    )
  }
  else {
    test <- NA
  }

  return(test)
}

getMeToCoffee("Indiana, PA", "driving", "dunkin") ## home
getMeToCoffee("100 Bureau Drive, Gaithersburg MD", "driving", "dunkin") ## NIST address
getMeToCoffee("Smith College, Northampton MA", "walking", "dunkin")
getMeToCoffee("Evans Hall, Berkeley CA", "driving", "dunkin")

Note: We actually do have a Dunkin in good-ole Indiana, PA so all is not lost.

# map("county","pennsylvania")
# points(coffee3$loc_LONG_centroid,coffee3$loc_LAT_centroid,col="orange")
# https://ryanpeek.github.io/2017-11-05-mapping-with-sf-Part-2/
# https://github.com/tidyverse/ggplot2/issues/2071

paDunkin <- subset(coffee3, e_state == "PA")
PA <- us_states(states = "pennsylvania")
ggplot() +
  geom_sf(data = PA, color = "black", fill = NA) +
  geom_point(data = paDunkin, aes(x = loc_LONG_centroid, y = loc_LAT_centroid)) +
  theme_void() +
  theme(panel.grid.major = element_line(colour = "white"))

Plenty

# map("county","maryland")
# points(coffee3$loc_LONG_centroid,coffee3$loc_LAT_centroid,col="orange")

mdDunkin <- subset(coffee3, e_state == "MD")
MD <- us_states(states = "maryland")
ggplot() +
  geom_sf(data = MD, color = "black", fill = NA) +
  geom_point(data = mdDunkin, aes(x = loc_LONG_centroid, y = loc_LAT_centroid)) +
  theme_void() +
  theme(panel.grid.major = element_line(colour = "white"))

Plenty

# map("county","massachusetts")
# points(coffee3$loc_LONG_centroid,coffee3$loc_LAT_centroid,col="orange")

maDunkin <- subset(coffee3, e_state == "MA")
MA <- us_states(states = "massachusetts")
ggplot() +
  geom_sf(data = MA, color = "black", fill = NA) +
  geom_point(data = maDunkin, aes(x = loc_LONG_centroid, y = loc_LAT_centroid)) +
  theme_void() +
  theme(panel.grid.major = element_line(colour = "white"))

Book me a plane ticket!

# map("county","california")
# points(coffee3$loc_LONG_centroid,coffee3$loc_LAT_centroid,col="orange")

caDunkin <- subset(coffee3, e_state == "CA")
CA <- us_states(states = "california")
ggplot() +
  geom_sf(data = CA, color = "black", fill = NA) +
  geom_point(data = caDunkin, aes(x = loc_LONG_centroid, y = loc_LAT_centroid)) +
  theme_void() +
  theme(panel.grid.major = element_line(colour = "white"))