reorder character class data frame column and plot - r

Question

I have a data frame that looks like the following:
df <- data.frame(date.time = c("Fri 00:00", "Fri 23:30", "Mon 00:00", "Mon 23:30",
"Sat 00:00", "Sat 23:30", "Sun 00:00", "Sun 23:30",
"Thu 00:00", "Thu 23:30", "Tue 00:00", "Tue 23:30",
"Wed 00:00", "Wed 23:30"),
Price = c(36.15368, 41.61206, 30.80412, 37.47360, 38.04516, 35.72798,
33.05613, 32.65447, 35.50335, 41.81241, 35.14006, 37.56432,
35.04553, 38.00721))
the date.time values are of class character and the Price values are of class numeric. I would like to plot the data using ggplot. The problem is that the data is in the wrong order. I would like an order of: sun, mon, ..., sat
I have attempted to do this using the following code:
my.order <- c(7,8,3,4,11,12,13,14,9,10,1,2,5,6)
df %>%
ggplot(aes(x = reorder(date.time, my.order), y = Price, group = 1)) +
geom_line()
but I end up getting a strange order that begins at the 'Tue' row of the original data frame. What am I doing wrong?
i would also like to label the x axis and so i have tried the following code:
df %>%
ggplot(aes(x = reorder(date.time, my.order), y = Price, group = 1)) +
geom_line() +
scale_x_discrete(name = 'Day', breaks = df$date.time[c(1,3,5,7,9,11,13)],
labels = c("Sun", "Mon", "Tue", "Wed", "Thu", "Fri", "Sat"))
But the labels end up in the order of the original data set, while the plot is ordered beginning on 'Tue' as above. How can I get both the data and labels to appear in the order I would like?
Edit: I think it might have something to do with the levels. Running the following code
df$date.time[c(7,8,3,4,11,12,13,14,9,10,1,2,5,6)]
results in the following output
[1] Sun 00:00 Sun 23:30 Mon 00:00 Mon 23:30 Tue 00:00 Tue 23:30 Wed 00:00 Wed 23:30
[9] Thu 00:00 Thu 23:30 Fri 00:00 Fri 23:30 Sat 00:00 Sat 23:30
14 Levels: Tue 00:00 Tue 23:30 Mon 00:00 Mon 23:30 Wed 00:00 Wed 23:30 ... Sun 23:30
Not sure why.

Answer 1

Your code actually does what you ask it to do in the first part of your problem: respecting the order of your data in df, you assigned position 1 and 2 to the two Tue values, which is why ggplot2 plots them first.
You can see the numbers associated to each element when running the following:
my.order <- c(7,8,3,4,11,12,13,14,9,10,1,2,5,6)
reorder(df$date.time, my.order)
You can use this vector for my.order instead:
my.order <- c(11,12,3,4,13,14,1,2,9,10,5,6,7,8)
df %>%
ggplot(aes(x = reorder(date.time, my.order), y = Price, group = 1)) +
geom_line()
The difference with the method df$date.time[c(7,8,3,4,11,12,13,14,9,10,1,2,5,6)] is that in your first reorder method, you associate a position to each element of your vector (i.e. 1st element has position 7, 2nd element has position 8, etc.) whereas, in the square bracket method you define the order in which elements in your vector come up (i.e. 7th element comes 1st, 8th element comes 2nd, etc.).
You will find that using the square bracket method in your ggplot call won't help as ggplot2 automatically uses the alphabetic order by default, i.e. the order of the data in your dataframe does not matter (the data being strings or factors won't make a difference).
However, if you use factors (which is the default when storing strings with the data.frame() function), you can order their levels:
df$date.time <- ordered(df$date.time,
levels = df$date.time[c(7,8,3,4,11,12,13,14,9,10,1,2,5,6)])
# see the new ordered levels
levels(df$date.time)
# visualise as is, ggplot2 uses ordered levels
df %>%
ggplot(aes(x = date.time, y = Price, group = 1)) +
geom_line()
For your labels, as the ordering of levels has not changed the order of your data in your dataframe, you still have to refer to their original position. But if you want your original code to work, you can add a step to reorganise your whole dataframe according to the ordered levels:
library(dplyr)
df <- df %>%
arrange(date.time)
The dplyr::arrange() function will take the ordered levels into account, and your rows are now ordered as expected.
Your original labelling method should then work fine:
df %>%
ggplot(aes(x = date.time, y = Price, group = 1)) +
geom_line() +
scale_x_discrete(name = 'Day', breaks = df$date.time[c(1,3,5,7,9,11,13)],
labels = c("Sun", "Mon", "Tue", "Wed", "Thu", "Fri", "Sat"))

Answer 2

To get Sunday to appear first do this:
df$date.time <- reorder(df$date.time, my.order)
df %>%
ggplot(aes(x = as.character(date.time), y = Price, group = 1)) +
geom_line()
No idea why, but making it a character sorts out the re-order issue.
EDIT: with as.character() it looks like the labels works as well?
df %>%
ggplot(aes(x = as.character(date.time), y = Price, group = 1)) +
geom_line() +
scale_x_discrete(name = 'Day', breaks = df$date.time[c(1,3,5,7,9,11,13)],
labels = c("Sun", "Mon", "Tue", "Wed", "Thu", "Fri", "Sat"))