covid_gif.R

library(data.table)
library(ggplot2)
library(gganimate)

population <- fread("/Users/chase/Downloads/csvData.csv") # https://worldpopulationreview.com/states
population <- population[, c("State", "Pop")]
colnames(population) <- c("state", "population")

covid <- "https://raw.githubusercontent.com/nytimes/covid-19-data/master/rolling-averages/us-states.csv"
covid <- fread(covid)


vaccinated <- "https://raw.githubusercontent.com/govex/COVID-19/master/data_tables/vaccine_data/us_data/time_series/people_vaccinated_us_timeline.csv"
vaccinated <- fread(vaccinated)

colnames(vaccinated)[which(colnames(vaccinated) == "Province_State")] <- "state"
colnames(vaccinated)[which(colnames(vaccinated) == "Date")] <- "date"

everything <- merge.data.table(covid, population, by = "state")
everything <- merge.data.table(everything, vaccinated, by = c("state","date"))

everything[, date := as.Date(everything$date)]

# NYTIMES data has pop-adjusted but to compare with vaccine data which isn't, we'll use the same population numbers for both
# everything[ , perc_full := (People_Fully_Vaccinated + People_Partially_Vaccinated)/population*100]
everything[ , perc_full := (People_Fully_Vaccinated)/population*100]
everything[, cases100 := cases_avg / population *100000 ] 

ggplot(everything,aes(x = perc_full, 
                      y = cases100)) +
  geom_point() +  
  xlab("Percent of State's Population That's Fully Vaccinated") +
  ylab("Covid Cases Per 100,000") +
  coord_cartesian( xlim=c(0, 100))+
  transition_time(date) +
  ease_aes('linear') + 
  theme(legend.position = "none") +
  labs(title = 'Date: {frame_time}') 

a <- ggplot(everything,aes(x = perc_full, 
                      y = cases_avg_per_100k)) +
  geom_text(aes(label=state)) +  
  xlab("Percent of State's Population That's Fully Vaccinated") +
  ylab("Covid Cases Per 100,000") +
  coord_cartesian( xlim=c(0, 100))+
  transition_time(date) +
  ease_aes('linear') + 
  theme(legend.position = "none") +
  labs(title = 'Date: {frame_time}') 

animate(a, 
        duration = 100,
        fps  =  3)

ggplot(everything[date == as.Date("2021-08-16")]) +
  geom_point(aes(x = perc_full, 
                 y = cases_avg_per_100k)) + 
  xlab("Percent of State's Population That's Fully Vaccinated") +
  ylab("Covid Cases Per 100,000") +
  coord_cartesian( xlim=c(0, 100)) +
  theme(legend.position = "none")