S Das S Das - 3 months ago 22
R Question

Order Factored Time Series Data

I have a dataset like the following:

a <- read.table(header=TRUE, text="Time Counter Speed
7:55:00 CT08W 63.79
9:30:00 CT08W 67.69
7:05:00 CT11W 68.34
8:40:00 CT10W 68.39
11:00:00 CT11W 68.00
10:40:00 CT01W 21.57
11:35:00 CT02W 72.03
5:25:00 CT12W 66.93
6:10:00 CT02W 62.79
12:05:00 CT02W 71.79 ")
a
Time Counter Speed
1 7:55:00 CT08W 63.79
2 9:30:00 CT08W 67.69
3 7:05:00 CT11W 68.34
4 8:40:00 CT10W 68.39
5 11:00:00 CT11W 68.00
6 10:40:00 CT01W 21.57
7 11:35:00 CT02W 72.03
8 5:25:00 CT12W 66.93
9 6:10:00 CT02W 62.79
10 12:05:00 CT02W 71.79

str(a)
'data.frame': 10 obs. of 3 variables:
$ Time : Factor w/ 10 levels "10:40:00","11:00:00",..: 8 10 7 9 2 1 3 5 6 4
$ Counter: Factor w/ 6 levels "CT01W","CT02W",..: 3 3 5 4 5 1 2 6 2 2
$ Speed : num 63.8 67.7 68.3 68.4 68 ...


I don't want to use
as.POSIXct
in the solution as it involves difficulties while annotating the figure.

ggplot(a, aes(x=Time, y=Counter, fill=Speed))


enter image description here

The plot shows that time is ordered wrongly. Is there a way to order the time while it is factored?

Answer

You can use the a date-time format for your time variable to order your factor.

The key is using the date-time variable to do the sorting, which can be done a variety of ways. Here, I create a date-time variable and sort it when making the factor variable.

a$dt = as.POSIXct(a$Time, format = "%H:%M:%S")

a$Time2 = factor(format(a$dt, "%H:%M:%S"), levels = format(sort(a$dt), "%H:%M:%S"))
a$Time2
[1] 07:55:00 09:30:00 07:05:00 08:40:00 11:00:00 10:40:00 11:35:00 05:25:00 06:10:00 12:05:00
Levels: 05:25:00 06:10:00 07:05:00 07:55:00 08:40:00 09:30:00 10:40:00 11:00:00 11:35:00 12:05:00

ggplot(a, aes(x=Time2, y=Counter, fill=Speed))

enter image description here

If using this method and the times aren't unique, you'll need to add unique into the levels argument.

Comments