bug_reports/issue_34.R

In OutbreakResources/incidence: Compute, Handle, Plot and Model Incidence of Dated Events

## Reproduce the error
library(incidence)
library(outbreaks)

onset <- ebola_sim$linelist$date_of_onset
i <- incidence(onset, interval = 7)
plot(i)

## Clue: this is not happening if we don't use the iso_week definition
i <- incidence(onset, 7, iso_week = FALSE)
plot(i)


## Explanation: the code generating iso weeks hacked through the ggplot2
## structure. Instead we specify our own labels and pass them through
## ggplot2::scale_x_date().

## This function takes a vector of Date objects, and an ideal number of breaks,
## and generates a list with two components: $breaks, and $labels. $breaks
## correspond to the first day of the matching iso week; $labels contains vector
## of labels of the corresponding iso weeks.

make_iso_weeks_breaks <- function(dates, n = 5) {
  breaks_ini <- pretty(dates, n)
  iso_weeks <- ISOweek::date2ISOweek(breaks_ini)
  iso_weeks_day1 <- sub("-[1-7]+$", "-1", iso_weeks)
  list(breaks = ISOweek::ISOweek2date(iso_weeks_day1),
       labels = sub("-[1-7]+$", "", iso_weeks)
       )
}

plot(i) + scale_x_date(breaks = breaks_day1, labels = labels)