how to plot year clock chart that resembles the ones by rhythm of food

I have the following data, which is just a part of more than 20,000 rows of data (you can import it into r by copying the following data and then run read.delim("clipboard",sep = ",")):

  "datetime","speed","month_name","year"
2011-01-01 01:00:00,10.14,"Jan",2011
2011-01-01 02:00:00,10.12,"Jan",2011
2011-01-01 03:00:00,8.94,"Jan",2011
2011-01-01 04:00:00,9.53,"Jan",2011
2011-01-01 05:00:00,11.68,"Jan",2011
2011-01-01 06:00:00,10.93,"Jan",2011
2011-01-01 07:00:00,11.01,"Jan",2011
2011-01-01 08:00:00,11.56,"Jan",2011
2011-01-01 09:00:00,10.47,"Jan",2011
2011-01-01 10:00:00,11.18,"Jan",2011
2011-01-01 11:00:00,10.82,"Jan",2011
2011-01-01 12:00:00,10.91,"Jan",2011
2011-01-01 13:00:00,11,"Jan",2011
2011-01-01 14:00:00,11.06,"Jan",2011
2011-01-01 15:00:00,11.95,"Jan",2011
2011-01-01 16:00:00,10.85,"Jan",2011
2011-01-01 17:00:00,11.22,"Jan",2011
2011-01-01 18:00:00,10.2,"Jan",2011
2011-01-01 19:00:00,11.43,"Jan",2011
2011-01-01 20:00:00,10.62,"Jan",2011
2011-01-01 21:00:00,10.26,"Jan",2011
2011-01-01 22:00:00,10.11,"Jan",2011
2011-01-01 23:00:00,9.81,"Jan",2011
2011-01-02 00:00:00,9.21,"Jan",2011
2011-01-02 01:00:00,8.7,"Jan",2011
2011-01-02 02:00:00,8.23,"Jan",2011
2011-01-02 03:00:00,8.98,"Jan",2011
2011-01-02 04:00:00,9.34,"Jan",2011
2011-01-02 05:00:00,8.57,"Jan",2011
2011-01-02 06:00:00,9.03,"Jan",2011

I would like to create a clock chart similar to the picture below, which is from google rhythm of food :

Here is my code and the output, which produces something completely different:

df %>% ggplot(aes(x = month_name,  y = speed)) +
  geom_bar(stat="identity", width = 0.5) +  
  coord_polar(start=0)

I want the radius of the circle to be the speed and the colour to be the frequency of that speed. similar to the following figure, but instead of directions, I want to have months around the circle the same as the first figure.

The code here doesn't work.

Solution

Your sample data isn't adequate to give a demonstration, so let's recreate a data set that has 3 years worth of hourly readings (which your own seems to have). The only variables we need are speed and datetime:

library(tidyverse)

set.seed(82)

df <- data.frame(datetime = seq(as.POSIXct("2011-01-01"), 
                                as.POSIXct("2014-01-01"), by = "hour"),
                 speed = abs(cumsum(rnorm(26305)))/2)

df %>%
  mutate(week  = lubridate::week(datetime),
         speed = cut(speed, breaks = seq(0, 40, 5)),
         speed = gsub("\\[|\\]|\\(|\\)", "", speed),
         speed = sapply(strsplit(speed, ","), 
                        function(x) mean(as.numeric(x)))) %>%
  count(week, speed) %>%
  ggplot(aes(x = week, y = speed)) +
  geom_point(data = expand.grid(week = 0:52 + 0.5, speed = seq(5, 40, 5)),
             color = "gray", size = 0.5) +
  geom_tile(aes(fill = n), width = 0.8, height = 4.5) +
  geomtextpath::geom_textsegment(
    data = data.frame(x = seq(0.5, 49.5, 52/12),
                      xend = seq(0.5, 49.5, 52/12) + 4,
                      label = toupper(month.abb)),
    inherit.aes = FALSE, color = "grey", vjust = -0.4, size = 8,
    aes(x = x, xend = xend, label = label, y = 45, yend = 45)) +
  coord_polar() +
  scale_fill_viridis_c(name = "Frequency") +
  scale_y_continuous(breaks = seq(0, 40, 5), limits = c(-10, 50)) +
  theme(panel.background = element_blank(),
        axis.text.x = element_blank(),
        axis.title.y = element_text(size = 16, hjust = 0.75),
        axis.title.x = element_blank(),
        legend.direction = "horizontal",
        legend.position = c(0.1, 0.1),
        legend.title = element_text(size = 16, vjust = 0.75))

^{Created on 2022-08-14 by the reprex package (v2.0.1)}