wikipedia scraper for population of TN cities

tn-city-wiki.R

library(dplyr)
library(foreach)
library(ggplot2)
library(ggrepel)
library(glue)
library(janitor)
library(rvest)

library(doParallel)
registerDoParallel(3)

cities <- c(
  "Knoxville", "Nashville", "Memphis", 
  "Chattanooga", "Clarksville", "Murfreesboro", 
  "Jackson", "Franklin"
)

get_city_population_data <- function(city) {
  tn <- read_html(glue("https://en.wikipedia.org/wiki/{city},_Tennessee"))
  
  tn_pop <- 
    tn %>% 
    html_nodes(".toccolours") %>% 
    html_table(header = T) %>% 
    `[[`(1) %>% 
    janitor::clean_names()
  
  names(tn_pop) <- tn_pop %>% slice(1) %>% as.character()
  
  tn_pop <- 
    tn_pop %>% 
    janitor::clean_names() %>% 
    slice(-c(1, nrow(tn_pop))) %>% 
    mutate(
      pop = gsub(",", "", pop) %>% as.numeric(),
      census = census %>% as.numeric()
      ) %>% 
    dplyr::select(census, pop)
  
  tn_pop
}

pops <- 
  foreach(i = seq_along(cities), .combine = rbind) %dopar% {
    get_city_population_data(cities[i]) %>% mutate(city = cities[i])
  }

pops %>% 
  mutate(label = if_else(census == (max(census) - 20), as.character(city), NA_character_)) %>%
  ggplot(aes(x = as.numeric(census), y = pop, color = city)) + 
  geom_point() + 
  geom_line() + 
  geom_label_repel(aes(label = label), force = 25, na.rm = TRUE, min.segment.length = 0) + 
  ggtitle("Population of major cities in Tennessee over time") + 
  scale_y_continuous(labels = scales::comma) + 
  xlab("Census Year") + 
  ylab("Population") + 
  scale_color_discrete(guide = FALSE) + 
  theme(text = element_text(size = 18))