rolling 28 day positive case counts in Scotland council areas using PHS open data and {cusumcharter}

scotland-cusums.R

library(dplyr)
library(data.table)
library(ggplot2)
library(cusumcharter)
library(purrr)
library(tidyr)
library(ggExtra)


# make the link dynamic
part1 <- "https://www.opendata.nhs.scot/dataset/"
part2 <- "b318bddf-a4dc-4262-971f-0ba329e09b87/"
part3 <- "resource/427f9a25-db22-4014-a3bc-893b68243055/"
part4 <- "download/trend_ca_"
part5 <- ".csv"
today <- gsub('-','',as.character(Sys.Date()))

link <- paste0(part1, part2, part3, part4, today, part5, sep = '')


dates <- seq.Date(as.Date(Sys.Date()-28), as.Date(Sys.Date()-1), by = '1 day')

DT <- data.table::fread(link)
DT[, Date := as.character(Date)]
DT[, Date := as.IDate(Date, format = "%Y%m%d")]
positives <- DT[Date >= as.Date(Sys.Date() -28),.(Date, CAName, DailyPositive)][]


# ggplot(positives,aes(Date, DailyPositive)) + 
#   geom_line() + 
#   geom_point() +
#   facet_wrap(~ CAName, ncol = 4, scales = "free_y") + 
#   theme_minimal() + 
#   labs(x = NULL, y = NULL) + 
#   ggExtra::rotateTextX()



p <- positives %>% 
  group_by(CAName) %>% 
  group_nest() %>% 
  mutate(cusums = purrr::map(.x = .$data, 
                             .f = ~ cusum_control_median(.x$DailyPositive))) %>% 
  tidyr::unnest( cols = c(data, cusums)) %>% 
  cusum_control_plot(.,
                     xvar = Date,
                     facet_var = CAName, 
                     facet_scales = 'free_y',
                     scale_type = "date",
                     datebreaks = "3 days",
                     title_text = "CUSUM Rolling 28 Day Positive Cases in Scottish Council Areas")

p <- p + facet_wrap( ~ CAName, ncol = 4, scales = 'free_y') +
  labs(subtitle = "Limits based on median over last 28 days") + 
  rotateTextX()
print(p) 

As at 2021-12-15