Hotspot drivers: All counties

rm(list = ls(all.names = TRUE))

library(readxl)
library(janitor)
library(dplyr)
library(tidyr)
library(lubridate)
library(forcats)
library(ggplot2)

master <- readxl::read_excel("master-mmwr-2.xlsx", sheet = "MASTER")

names(master)

 [1] "Region"        "Week"          "Report_Date"   "Data_Date"     "County"        "State"        
 [7] "Pop"           "FIPS"          "New_Status"    "Tribal"        "Driver_1"      "Driver_2"     
[13] "Driver_3"      "Driver_4"      "Driver_5"      "Drivers"       "Demographics"  "Cases"        
[19] "Testing"       "Location"      "Prev_Report"   "Narrative"     "Outreach_Date" "Outreach_Team"
[25] "Outreach_Type"

# combine all driver categories and count them
long <- dplyr::select(master, Region:Driver_5) %>%
  tidyr::pivot_longer(Driver_1:Driver_5) %>%
  dplyr::mutate(week_num = lubridate::week(Data_Date))

table(master$Driver_1, useNA = "always")

   C    D    F    L    M  n/a    P    S    T   TR    U    W <NA> 
2638   49  247  451   43  256  275    4  340    9 1123   88    0 

# count of each driver by FIPS
t1 <-  long %>%
  dplyr::group_by(Region, County, State, FIPS) %>%
  dplyr::count(value) %>%
  tidyr::drop_na() %>%
  tidyr::pivot_wider(names_from = value, values_from = n)
t1

# count number of days listed per county
day_counts_fips <- long %>%
  dplyr::group_by(Region, County, State, FIPS) %>%
  dplyr::summarise(ndays = n())

`summarise()` regrouping output by 'Region', 'County', 'State' (override with `.groups` argument)

day_counts_fips

# these two values should equal; if they don't then:
# there are rows without any driver information, and/or
# there are FIPS code where no driver is listed
nrow(t1)

[1] 656

nrow(day_counts_fips)

[1] 656

# add column
driver_and_day <- dplyr::full_join(t1, day_counts_fips, by = c("Region", "County", "State", "FIPS")) %>%
  dplyr::select(Region:FIPS, ndays, everything()) # reorder columns
driver_and_day

# plot county histogram
p_hist <- ggplot(driver_and_day) +
  geom_bar(mapping = aes(x = ndays))
p_hist

# facet histogram by region
p_hist_region <- p_hist + facet_wrap(~Region)
p_hist_region

# count drivers by week
group_by_week <- long %>%
  dplyr::group_by(week_num) %>%
  dplyr::count(value) %>%
  tidyr::drop_na() %>%
  dplyr::filter(value != "n/a")
group_by_week

# set up plot
g <- ggplot(group_by_week,
            aes(x = as.factor(week_num),
                y = n,
                fill = forcats::fct_reorder(value, n))) +
  theme_minimal()

# stacked bar chart by count
g + geom_bar(stat = "identity", position = "stack")

#ggsave("fig1.pdf")

# bar chart
g + geom_bar(stat = "identity", position = "dodge")

#ggsave("fig2.pdf")

# stacked bar chart by proportion 
g + geom_bar(stat = "identity", position = "fill")

#ggsave("fig3.pdf")

# line graph of number of each driver over time
ggplot(group_by_week, aes(x = week_num, y = n, color = value)) + 
  geom_line() +
  theme_minimal()

#ggsave("fig4.pdf")

# group by region and week
group_by_region_week <- long %>%
  dplyr::group_by(Region, week_num) %>%
  dplyr::count(value) %>%
  tidyr::drop_na() %>%
  dplyr::filter(value != "n/a")
group_by_region_week

# line graph of each driver over time facetted by region
ggplot(group_by_region_week,
       aes(x = week_num,
           y = n,
           color = value)) + 
  geom_line() +
  facet_wrap(~Region) +
  theme_minimal() +
  ggtitle("Frequency of drivers over time by region")

#ggsave("fig5.pdf")

# plot regions over time facetted by driver
ggplot(group_by_region_week,
       aes(x = week_num,
           y = n,
           color = as.factor(Region))) +
  geom_line() +
  facet_wrap(~value) +
  theme_minimal() +
  ggtitle("Driver trend by region")

#ggsave("fig6.pdf")

# convert the above code into readable table
group_by_region_week %>%
  tidyr::pivot_wider(names_from = value, values_from = n)