library(rgdal)
library(tidyverse)
library(scales)

la <- readOGR(paste0("Local_Authority_Districts_December_2015_Super_Generalised_Clipped_",
"Boundaries_in_Great_Britain/Local_Authority_Districts_December_2015_Super_Generalised_Clipped_",
              "Boundaries_in_Great_Britain.shp"))

# Change the coordinate reference system - it's a bit easier to work with
la <- spTransform(la, CRS("+init=epsg:4326"))

# spdf contain @data and @polygons (@ = S4 objects). The rownames() in @data correspond
# to the IDs in @polygons (yeah, it's fucked up)
la@data$id <- rownames(la@data)

# fortify() is being deprecated - use broom::tidy() instead
la_df <- fortify(la)
la_df <- inner_join(la_df, la@data, by = "id")

# Data import and tidying
med_house_price <- read_csv("med_house_price.csv")

# Transform to long format
med_house_price <- gather(med_house_price, period, house_price,
                          -`Region code`, -`Region name`, -`Local authority code`, -`Local authority name`)
# Change to years
med_house_price$period <- substr(med_house_price$period, 4, 8)

med_gross_income <- read_csv("med_gross_income.csv")
med_gross_income <- gather(med_gross_income, period, median_salary,
                           -`Region code`, -`Region name`, -`Local authority code`, -`Local authority name`)

# Reformat totally fucked salary format
med_gross_income <- med_gross_income %>% mutate(median_salary = median_salary %>% map_int(function(x) {
  out <- gsub(",", "", x)
  out <- gsub(":", "", out)
  as.integer(out)
}))

price_salary_ratio <- read_csv("price_salary_ratio.csv")
price_salary_ratio <- gather(price_salary_ratio, period, price_salary_ratio,
                             -`Region code`, -`Region name`, -`Local authority code`, -`Local authority name`)

# Reformat totally fucked price salary ratio format
price_salary_ratio <- price_salary_ratio %>% mutate(price_salary_ratio = price_salary_ratio %>% map_dbl(function(x) {
  out <- gsub(",", "", x)
  out <- gsub(":", "", out)
  out <- as.double(out)
  out <- ifelse(out >= 15, 15, out)
}))

price_salary_ratio <- inner_join(med_gross_income, price_salary_ratio)
price_salary_ratio <- inner_join(med_house_price, price_salary_ratio)
price_salary_ratio$authority <- price_salary_ratio$`Local authority code`
la_df$lad15cd <- as.character(la_df$lad15cd)

# Combine our data and spatial data frame
la_df <- inner_join(la_df, price_salary_ratio, by = c("lad15cd" = "authority"))

ggplot(la_df, aes(long, lat)) +
  geom_polygon(aes(group = group, fill = price_salary_ratio), color = "white", size = 0.2) +
  coord_quickmap() +
  scale_fill_gradient2(low = "blue4", mid = "white", high = "red4",
                       midpoint = quantile(price_salary_ratio$price_salary_ratio, na.rm = TRUE,
                                           probs = 0.5)) +
  theme_bw() +
  labs(title = "Housing Affordability - England & Wales 2016",
       subtitle = "The ratio of median house price and yearly gross earnings per household",
       fill = "Price / Salary \nRatio\n") +
  theme(panel.background = element_rect(fill = 'grey99', colour = 'grey99'),
        axis.title.x = element_blank(),
        axis.title.y = element_blank())

la_df_london <- filter(la_df, `Region name` == "London")
ggplot(la_df_london, aes(long, lat)) +
  geom_polygon(aes(group = group, fill = price_salary_ratio), color = "grey10", size = 0.2) +
  coord_quickmap() +
  scale_fill_gradient2(low = "blue4", mid = "white", high = "red4",
                       midpoint = quantile(price_salary_ratio$price_salary_ratio, na.rm = TRUE,
                                           probs = 0.5)) +
  theme_bw() +
  labs(title = "Housing Affordability - England & Wales 2016",
       subtitle = "The ratio of median house price and yearly gross earnings per household",
       fill = "Price / Salary \nRatio\n") +
  theme(panel.background = element_rect(fill = 'grey99'),
        axis.title.x = element_blank(),
        axis.title.y = element_blank(),
        axis.text.x = element_blank(),
        axis.text.y = element_blank())