Apple Mobility and EU CDC cases

library(dplyr)
library(countrycode)
library(lubridate)
library(tidyr)
library(ggplot2)
library(ggthemes)
library(ggrepel)
# apple mobility data csv from
# https://www.apple.com/covid19/mobility
amt <- read.csv("~/Desktop/applemobilitytrends-2020-05-20.csv", colClasses = "character",
                check.names = FALSE)
data_extent <- ncol(amt)

# EU CDC
# https://www.ecdc.europa.eu/en/geographical-distribution-2019-ncov-cases
euCDC <- read.csv("https://opendata.ecdc.europa.eu/covid19/casedistribution/csv", 
                  stringsAsFactors = FALSE)

depth <- amt %>%
  filter(transportation_type == "driving", geo_type=="country/region") %>%
  gather(key="Date", value="mobility", 7:data_extent) %>% 
  select(region, Date, transportation_type, mobility) %>%
  mutate(Date = ymd(Date), mobility=as.numeric(mobility)) %>%
  arrange(region, Date) %>%
  group_by(region) %>%
  mutate(sevendaymob = mobility + lag(mobility, 1) + lag(mobility, 2) + lag(mobility, 3) +
           lag(mobility, 4) + lag(mobility, 5) + lag(mobility, 6)) %>% 
  summarise(depth = 100* min(sevendaymob, na.rm=T)/max(sevendaymob, na.rm=T)) %>%
  arrange(depth)

apple <- amt %>%
  filter(transportation_type == "driving", geo_type=="country/region") %>%
  gather(key="Date", value="mobility", 7:data_extent) %>% 
  select(region, Date, transportation_type, mobility) %>%
  mutate(Date = ymd(Date), mobility=as.numeric(mobility)) %>%
  arrange(region, Date) %>%
  group_by(region) %>%
  mutate(sevendaymob = (mobility + lag(mobility, 1) + lag(mobility, 2) + lag(mobility, 3) +
           lag(mobility, 4) + lag(mobility, 5) + lag(mobility, 6))/7) %>%
  ungroup() %>%
  filter(sevendaymob <60) %>%
  arrange(region, Date) %>%
  group_by(region) %>%
  slice(1) %>%
  ungroup() %>% select(region,speed = Date) %>% inner_join(depth) %>%
  mutate(ccode = countrycode(region, 'country.name', 'iso3c'))
euCDC %>%
  mutate(Date30 = dmy(dateRep)) %>%
  arrange(countryterritoryCode, Date30) %>%
  group_by(countryterritoryCode) %>%
  mutate(total_cases = cumsum(cases)) %>%
  ungroup() %>%
  filter(total_cases >= 30) %>%
  group_by(countryterritoryCode) %>%
  slice(1) %>%
  ungroup() %>%
  select(ccode= countryterritoryCode, Date30) %>%
  inner_join(apple) %>% 
  mutate(days_from_thirty = as.numeric(difftime(speed, Date30, units = "days"))) %>% 
  ggplot(aes(x=days_from_thirty, y=depth, label=region)) + 
  geom_point() + geom_label_repel() + theme_minimal() + ylab("Depth of driving immobility (percent min week/max week)") +
  xlab("Speed of Lockdown (day of end of week after 30th case driving falls below 60% for week)") +
  ggtitle("Speed (relative to 30th case) and strength (reduction in driving) of Lockdowns")+
  labs(caption="data from EUCDC & Apple")