Plots and gives some info on correlation

cases_to_deaths.R

library(tidyverse)
library(lubridate)
library(zoo)
library(scales)

# Read Data ---------------------------------------------------------------


confirmed <- read_csv("https://github.com/nytimes/covid-19-data/raw/master/us-states.csv")


# Get New Deaths + Cases --------------------------------------------------

df <- confirmed
df$new_deaths <- df$deaths - df$deaths[c(1,1:(length(df$deaths)-1))]
df$new_cases <- df$cases - df$cases[c(1,1:(length(df$cases)-1))]
df$date <- ymd(df$date)

# Filtering out where new_cases is less than 0
df_dates <- data.frame(
  date = seq(ymd("2021-01-01"),max(df$date),1)
) %>% 
  left_join(df %>% filter(new_cases>=0)) %>%
  mutate(
    roll_mean_deaths = new_deaths %>% rollmean(7,fill = 0, align = "right"),
    roll_mean_cases = new_cases %>% rollmean(7,fill = 0, align = "right")
  ) 

# Set Offset
offset <- 12
df_dates$offset_deaths <- df_dates$roll_mean_deaths
df_dates$offset_deaths[1:(nrow(df_dates)-offset)] <- df_dates$offset_deaths[((offset+1):nrow(df_dates))+offset]
df_dates$offset_deaths[(nrow(df_dates)-(offset-1)):nrow(df_dates)] <- NA

# Plot --------------------------------------------------------------------

ggplot(df_dates,aes(x=roll_mean_cases,y=offset_deaths)) + geom_point() + ggtitle(offset) + geom_smooth()
lm(offset_deaths~roll_mean_cases,data = df_dates)%>% summary

for(offset in 10:20) {
  df_dates$offset_deaths <- df_dates$roll_mean_deaths
  df_dates$offset_deaths[1:(nrow(df_dates)-offset)] <- df_dates$offset_deaths[((offset+1):nrow(df_dates))+offset]
  df_dates$offset_deaths[(nrow(df_dates)-(offset-1)):nrow(df_dates)] <- NA
  # Plot --------------------------------------------------------------------
  p1 <- ggplot(df_dates,aes(x=roll_mean_cases,y=offset_deaths)) + geom_point() + ggtitle(offset) + geom_smooth()
  print(p1)
}