top_metros.R

library(ggplot2)
library(animation)
library(readxl)
library(tidyr)
library(dplyr)
library(stringr)
library(magrittr)

# Data source: Jonathan Schroeder at http://conservancy.umn.edu/handle/11299/181605
df <- read_excel('cbsa2013_hist_pops.xlsx')

dft <- df %>%
  gather(key = year, value = population, epop1790:pop2010) %>%
  mutate(year = as.numeric(str_sub(year, start = -4)))

dft$name_short <- str_split_fixed(dft$CBSA_NAME, ",", n = 2)[,1]

dft$name_short <- str_split_fixed(dft$name_short, "-", n = 2)[,1]

dft$name_short <- str_split_fixed(dft$name_short, "/", n = 2)[,1]

top20 <- dft %>%
  group_by(year) %>%
  mutate(rank = min_rank(desc(population))) %>%
  ungroup() %>%
  filter(rank <= 20)

northeast <- c("Albany", "Boston", "Bridgeport", "Claremont", "Hartford", "New York", 
               "Philadelphia", "Pittsburgh", "Portland", "Providence", "Springfield", 
               "Torrington", "Worcester", "Rochester", "Syracuse", "Buffalo", "Scranton")

midwest <- c("Cincinnati", "Columbus", "Chicago", "St. Louis", "Indianapolis", "Detroit", 
             "Kansas City", "Cleveland", "Minneapolis", "Milwaukee")

south <- c("Baltimore", "Charleston", "Charlotte", "Richmond", "Salisbury", "Virginia Beach", "Washington", 
           "Charlottesville", "Lexington", "Nashville", "Louisville", "Atlanta", "New Orleans", 
           "Dallas", "Houston", "Miami", "Tampa")

west <- c("San Francisco", "Los Angeles", "Seattle", "San Diego", "Phoenix", "Riverside")

top20 %<>%
  mutate(region = ifelse(name_short %in% northeast, "Northeast", 
                                ifelse(name_short %in% midwest, "Midwest", 
                                       ifelse(name_short %in% south, "South", 
                                              ifelse(name_short %in% west, "West", NA)))), 
         poplabel = ifelse(population < 1000000, paste0(as.character(round(population / 1000, 0)), "k"), 
                           paste0(as.character(round(population / 1000000, 2)), "m")), 
         position_label = ifelse(population < 100000, 1000000, 1200000), 
         name_short = str_pad(top20$name_short, 15, "left"))

saveGIF({
  
  for (i in seq(1790, 2010, 10)) {
    
    yearly <- filter(top20, year == i)
    
    g <- ggplot() + 
      geom_bar(data = yearly, aes(y = population, x = reorder(name_short, population), 
                                 fill = region, frame = year), stat = "identity") + 
      geom_text(data = yearly, aes(y = population + position_label, x = reorder(name_short, population), 
                                  label = poplabel), fontface = "bold") + 
      coord_flip() +
      scale_y_continuous(limits = c(0, 25000000), expand = c(0, 0)) + 
      scale_fill_manual(values = c("Northeast" = "#e41a1c", "Midwest" = "#377eb8", 
                                    "South" = "#4daf4a", "West" = "#984ea3")) + 
      theme_minimal(base_size = 16, base_family = "Tahoma") + 
      theme(panel.grid.major.x = element_blank(), 
            panel.grid.minor.x = element_blank(), 
            panel.grid.major.y = element_blank(), 
            panel.grid.minor.y = element_blank(), 
            plot.title = element_text(face = "bold"), 
            axis.text.x = element_blank(), 
            legend.position = "bottom") + 
      labs(y = "", 
           x = "", 
           fill = "", 
           caption = "Data source: Jonathan Schroeder, University of Minnesota | Chart by @kyle_e_walker", 
           title = paste0("20 largest US metro areas by population, ", as.character(i))) 
    
    print(g)
    
  }
  
  
}, movie.name = "metro_pop.gif", interval = 0.8, ani.width = 700, ani.height = 600)