R Notebook

library(tidyverse)
library(timetk)

Pull data and process

raw_covid19_tbl <- read_csv("https://raw.githubusercontent.com/nytimes/covid-19-data/master/us-counties.csv")


data_daily <- raw_covid19_tbl %>% 
  filter(state%in%c("Illinois", "Indiana", #Pull Midwestern states. 
                    "Iowa", "Wisconsin","Missouri")) %>% 
  group_by(state,date) %>% 
  summarise(cases=sum(cases),
            deaths=sum(deaths)) %>% 
  ungroup() %>%
  group_by(state) %>% 
  arrange(date,.by_group = TRUE) %>%
  mutate(cases_new = cases - lag(cases)) %>% #Compute daily deltas
  mutate(deaths_new = deaths - lag(deaths))  #Compute daily deltas

data_daily

Enter timetk Package

onvert data to weekly and summarize cases and deaths.

data_weekly <-data_daily %>% 
  summarise_by_time(.date_var=date, 
                    .by="week",
                    cases_new=sum(cases_new),
                    deaths_new=sum(deaths_new)) 


data_weekly

NA

data_weekly %>% 
  plot_time_series(date,cases_new, 
                   .smooth_period = "2 weeks",
                   .smooth_span=0.75,
                   .smooth_degree =1,
                   .smooth_message= T,
                   .facet_ncol = 3)

# try to facet the plot 

data_weekly %>% 
  plot_acf_diagnostics(date, deaths_new, .lags="3 months")

data_weekly %>%  
  drop_na() %>% 
  filter(state=="Illinois") %>% 
  plot_anomaly_diagnostics(date, deaths_new)

frequency = 11.5 observations per 1 quarter
trend = 13 observations per 3 months