How to plot time series data as boxplot in a time unit?

Question

I have the following test time-series data. I'd like to plot a daily/weekly/monthly boxplot. But it doesn't do so. How can I fix the problem?

library(ggplot2)
library(dplyr)
data <- data.frame(
  day = rep(as.Date("2017-06-14") - 0:364, 100),
  value = unlist(replicate(100, list(rnorm(365) + seq(-140, 224)^2 / 10000)))
)

p = ggplot(data, aes(group=day, y=value)) + geom_boxplot() + scale_x_date(date_breaks = "1 week", date_labels = "%W")

Answer 1

We can use ceiling_date from lubridate, and use that in geom_boxplot

library(dplyr)
library(ggplot2)
library(lubridate)
data  %>% 
    mutate(week = ceiling_date(day, "week")) %>%
    ggplot(aes(group = week, y = value)) + 
    geom_boxplot() + 
    scale_x_date(date_breaks = "1 week", date_labels = "%W")

---

Or it could be

data  %>%
      mutate(week = format(day, "%W")) %>%
      ggplot(aes(x = week, y = value, fill = factor(week))) + 
        geom_boxplot() + 
        theme_bw() + 
        theme(legend.position = "none",
           axis.text.x = element_text(angle = 90,  hjust=1))