Delta vs. Omicron … some numbers – Minimum Viable Python

Let’s do two COVID-19 plots for Los Angeles county cases over the past 6 months. Let’s assume we have a file of te daily reports from http://publichealth.lacounty.gov right here in this directory!

First, read the data

Code

library(dplyr)
library(lubridate) # for working with dates
library(ggplot2)  # for creating plots of the surge


### SET UP Caption and title
captionSource='data source: daily reports at http://publichealth.lacounty.gov'
titlestring <- paste('LA County new Covid19 cases modeled thru ',today()-1)


cases <- read.csv("LACountyNewDailyCases.csv")
# Clean the bad input data!
#cases$mnum <- plyr::revalue(cases$Month, c("Jan"="01", "jan"="01", "Dec"=12, "dec"=12, "Feb"="02","Mar"="03","Apr"="04","may"="05", "May"="05","Jun"="06"))

## Made some typos in the input file.  Will it happen again?
### (It will happen again!)
cases$mnum <- dplyr::recode(cases$Month, "Jan"="01", "jan"="01", "Dec"="12", "dec"="12", "Feb"="02","Mar"="03","Apr"="04","may"="05", "May"="05","Jun"="06")

# makin a uniform date from "why did he do that?" ymd columns
cases$dstring <- paste0(cases$Year,cases$mnum,cases$Day)
cases$date <- ymd(cases$dstring)

So now, add a column showing number of cases per 100 thousand for each date. And, look at the most recent cases

Code

LACountyPop = 10007445
Divisor100K <- LACountyPop / 100000
cases$per100K <- cases$Cases / Divisor100K

#LA county is ten million, so ... rate per 100K is eyeball-able from the case rate, too

head(cases)

  Year Month Day Cases mnum  dstring       date   per100K
1 2023   Jan  17   510   01 20230117 2023-01-17  5.096206
2 2023   Jan  16   477   01 20230116 2023-01-16  4.766451
3 2023   Jan  15   809   01 20230115 2023-01-15  8.083981
4 2023   Jan  14  1421   01 20230114 2023-01-14 14.199429
5 2023   Jan  13  1534   01 20230113 2023-01-13 15.328588
6 2023   Jan  12  1716   01 20230112 2023-01-12 17.147234

Send the data to a plot showing the increase and decrease of the case rate as the Delta variant hits Los Angeles County in December of 2021

Code

cases |> ggplot(aes(x=date,y=Cases)) + geom_line() +
  labs(title=titlestring,
       subtitle = "vertical lines show Christmas,\n New Years,\n MLK Day,\n SuperBowl\nhorizontal redline is 1500",
       x="Date",
       y= "cases",
       caption=captionSource) +
  geom_vline(xintercept = as.numeric(as.Date("2021-12-25")), linetype=4, color='red')  +
  geom_vline(xintercept = as.numeric(as.Date("2022-01-01")), linetype=4, color='green') +
  geom_vline(xintercept = as.numeric(as.Date("2022-01-17")), linetype=10, color='red') +
  geom_vline(xintercept = as.numeric(as.Date("2022-02-13")), linetype=9, color='red') +    geom_hline(yintercept = 1500, color='red')

Now .. add a column named ‘proj’ (projected), assuming there are 6 actual cases ofr every reported case

Code

checkpoint <- date('2022-04-23')
aftercheck <- interval(checkpoint,today())

# TESTING: cases$date %within% aftercheck

# 5 or 6, right?
mulfactor <- 6

## Wrong!
# ifelse(cases$date %within% aftercheck,
#   cases$proj <- mulfactor * cases$Cases,
#   cases$proj <- cases$Cases
# )

## Right!
cases$proj <- ifelse(cases$date %within% aftercheck,
       mulfactor * cases$Cases,
       cases$Cases
)


## TESTING: cases$proj == cases$Cases

## OK

Now, graph that modeled current-surge (but leave the Delta variant curve the same)

Code

## Now, plotting 'proj'

cases |> ggplot(aes(x=date,y=proj)) + geom_line() +
  labs(title=titlestring,
       subtitle = "vertical lines show Christmas,\n New Years,\n MLK Day,\n SuperBowl\nhorizontal orange line is 8000",
       x="Date",
       y= "cases, except cases times N=5 or N=6 after April 23",
       caption=captionSource) +
  geom_vline(xintercept = as.numeric(as.Date("2021-12-25")), linetype=4, color='red')  +
  geom_vline(xintercept = as.numeric(as.Date("2022-01-01")), linetype=4, color='green') +
  geom_vline(xintercept = as.numeric(as.Date("2022-01-17")), linetype=10, color='red') +
  geom_vline(xintercept = as.numeric(as.Date("2022-02-13")), linetype=9, color='red') +    
  geom_hline(yintercept = 8000, color='orange')

Note: you can see what happened here after I wrote about this on June 2 .. I kept following the data, and just moved the updated file over and re-rendered in Quarto.

I feel like this is a time when a lot of people are getting COVID and surviving .. but the hospitalizations – which ARE reported at publichealth.lacounty.gov – are not where they were back in the beginning of the year.

I do like the non-‘modeled’ sequence better, because otherwise you have to guess at a differential rate of reporting which, in the toy example case above, is just invented.

Code

## Now, plotting 'proj'
cases |> ggplot(aes(x=date,y=Cases)) + geom_col() +
  labs(title=titlestring,
       subtitle = "lines show Christmas, New Years, MLK Day, superbowl/valentines\nhorizontal is 1500",
       x="Date",
       y= "cases",
       caption=captionSource) +
  geom_vline(xintercept = as.numeric(as.Date("2021-12-25")), linetype=4, color='red')  +
  geom_vline(xintercept = as.numeric(as.Date("2022-01-01")), linetype=4, color='green') +
  geom_vline(xintercept = as.numeric(as.Date("2022-01-17")), linetype=4, color='yellow') +
  geom_vline(xintercept = as.numeric(as.Date("2022-02-13")), linetype=9, color='red') +    geom_hline(yintercept = 1500, color='red')

--- title: "Delta vs. Omicron ... some numbers" author: "George Girton" date: "2022-06-02" categories: [code, analysis] image: "IMG_0887_sweb.png" --- Let's do two COVID-19 plots for [Los Angeles county cases](LACountyNewDailyCases.csv) over the past 6 months. Let's assume we have a file of te daily reports from http://publichealth.lacounty.gov right here in this directory! First, read the data ```{r } #| echo: true #| fig-cap: "Calculated multiplier after Apr 23 data" #| warning: false #| library(dplyr) library(lubridate) # for working with dates library(ggplot2) # for creating plots of the surge ### SET UP Caption and title captionSource='data source: daily reports at http://publichealth.lacounty.gov' titlestring <- paste('LA County new Covid19 cases modeled thru ',today()-1) cases <- read.csv("LACountyNewDailyCases.csv") # Clean the bad input data! #cases$mnum <- plyr::revalue(cases$Month, c("Jan"="01", "jan"="01", "Dec"=12, "dec"=12, "Feb"="02","Mar"="03","Apr"="04","may"="05", "May"="05","Jun"="06")) ## Made some typos in the input file. Will it happen again? ### (It will happen again!) cases$mnum <- dplyr::recode(cases$Month, "Jan"="01", "jan"="01", "Dec"="12", "dec"="12", "Feb"="02","Mar"="03","Apr"="04","may"="05", "May"="05","Jun"="06") # makin a uniform date from "why did he do that?" ymd columns cases$dstring <- paste0(cases$Year,cases$mnum,cases$Day) cases$date <- ymd(cases$dstring) ``` So now, add a column showing number of cases per 100 thousand for each date. And, look at the most recent cases ```{r } LACountyPop = 10007445 Divisor100K <- LACountyPop / 100000 cases$per100K <- cases$Cases / Divisor100K #LA county is ten million, so ... rate per 100K is eyeball-able from the case rate, too head(cases) ``` Send the data to a plot showing the increase and decrease of the case rate as the Delta variant hits Los Angeles County in December of 2021 ```{r } #| echo: true #| fig-cap: "Straight up data" #| warning: false cases |> ggplot(aes(x=date,y=Cases)) + geom_line() + labs(title=titlestring, subtitle = "vertical lines show Christmas,\n New Years,\n MLK Day,\n SuperBowl\nhorizontal redline is 1500", x="Date", y= "cases", caption=captionSource) + geom_vline(xintercept = as.numeric(as.Date("2021-12-25")), linetype=4, color='red') + geom_vline(xintercept = as.numeric(as.Date("2022-01-01")), linetype=4, color='green') + geom_vline(xintercept = as.numeric(as.Date("2022-01-17")), linetype=10, color='red') + geom_vline(xintercept = as.numeric(as.Date("2022-02-13")), linetype=9, color='red') + geom_hline(yintercept = 1500, color='red') ``` Now .. add a column named 'proj' (projected), assuming there are 6 actual cases ofr every reported case ```{r} #| echo: true #| warning: false #| checkpoint <- date('2022-04-23') aftercheck <- interval(checkpoint,today()) # TESTING: cases$date %within% aftercheck # 5 or 6, right? mulfactor <- 6 ## Wrong! # ifelse(cases$date %within% aftercheck, # cases$proj <- mulfactor * cases$Cases, # cases$proj <- cases$Cases # ) ## Right! cases$proj <- ifelse(cases$date %within% aftercheck, mulfactor * cases$Cases, cases$Cases ) ## TESTING: cases$proj == cases$Cases ## OK ``` Now, graph that modeled current-surge (but leave the Delta variant curve the same) ```{r} #| echo: true #| fig-cap: "Model-scaled data after Apr23" #| warning: false #| ## Now, plotting 'proj' cases |> ggplot(aes(x=date,y=proj)) + geom_line() + labs(title=titlestring, subtitle = "vertical lines show Christmas,\n New Years,\n MLK Day,\n SuperBowl\nhorizontal orange line is 8000", x="Date", y= "cases, except cases times N=5 or N=6 after April 23", caption=captionSource) + geom_vline(xintercept = as.numeric(as.Date("2021-12-25")), linetype=4, color='red') + geom_vline(xintercept = as.numeric(as.Date("2022-01-01")), linetype=4, color='green') + geom_vline(xintercept = as.numeric(as.Date("2022-01-17")), linetype=10, color='red') + geom_vline(xintercept = as.numeric(as.Date("2022-02-13")), linetype=9, color='red') + geom_hline(yintercept = 8000, color='orange') ``` Note: you can see what happened here after I wrote about this on June 2 .. I kept following the data, and just moved the updated file over and re-rendered in Quarto. I feel like this is a time when a lot of people are getting COVID and surviving .. but the hospitalizations -- which ARE reported at publichealth.lacounty.gov -- are not where they were back in the beginning of the year. ---- I do like the non-'modeled' sequence better, because otherwise you have to guess at a differential rate of reporting which, in the toy example case above, is just invented. ```{r} #| echo: true #| fig-cap: "bar chart of cases" #| warning: false #| ## Now, plotting 'proj' cases |> ggplot(aes(x=date,y=Cases)) + geom_col() + labs(title=titlestring, subtitle = "lines show Christmas, New Years, MLK Day, superbowl/valentines\nhorizontal is 1500", x="Date", y= "cases", caption=captionSource) + geom_vline(xintercept = as.numeric(as.Date("2021-12-25")), linetype=4, color='red') + geom_vline(xintercept = as.numeric(as.Date("2022-01-01")), linetype=4, color='green') + geom_vline(xintercept = as.numeric(as.Date("2022-01-17")), linetype=4, color='yellow') + geom_vline(xintercept = as.numeric(as.Date("2022-02-13")), linetype=9, color='red') + geom_hline(yintercept = 1500, color='red') ``` ----