R: Superimposing time series

Question

So the idea is simple, and I guess the solution lies somewhere, but I just cannot find it. I have 3 yearly time series, with daily values, and I'd like to be able to display all 3 years in the same graph, where the x coordinates are the months, and the y coordinates the values for each year.

Answer 1

I would suggest a ggplot2 approach to your consideration using dummy data. If you want months in x-axis you can 'hack' the date assigning a new year so that you can see the months in the axis after formating with scale_x_date(). You also have to exatract the years from your date in order to add the colors. Here the code:

library(ggplot2)
library(dplyr)
set.seed(123)
#Dummy data
df <- data.frame(Date=c(seq(as.Date('2017-01-01'),
                          as.Date('2017-12-31'),
                          length.out = 365),
                        seq(as.Date('2018-01-01'),
                            as.Date('2018-12-31'),
                            length.out = 365),
                        seq(as.Date('2019-01-01'),
                            as.Date('2019-12-31'),
                            length.out = 365)),
                 value=rnorm(1095,5,2))
#Format
#Create months and years
df %>% mutate(Year=format(Date,'%Y'),
              Date2=as.Date(paste0('2020-',format(Date,'%m-%d')))) %>%
  ggplot(aes(x=Date2,y=value,group=Year,color=Year))+
  geom_line()+
  scale_x_date(date_labels="%B",breaks = '1 month')

Output:

Update: Also credits to great @AllanCameron, here a nicer solution using cumsum() for the variable value:

#Dummy data
df <- data.frame(Date=c(seq(as.Date('2017-01-01'),
                          as.Date('2017-12-31'),
                          length.out = 365),
                        seq(as.Date('2018-01-01'),
                            as.Date('2018-12-31'),
                            length.out = 365),
                        seq(as.Date('2019-01-01'),
                            as.Date('2019-12-31'),
                            length.out = 365)),
                 value = c(replicate(3, cumsum(rnorm(365)))))
#Format
#Create months and years
df %>% mutate(Year=format(Date,'%Y'),
              Date2=as.Date(paste0('2020-',format(Date,'%m-%d')))) %>%
  ggplot(aes(x=Date2,y=value,group=Year,color=Year))+
  geom_line()+
  scale_x_date(date_labels="%B",breaks = '1 month')

Output:

Answer 2

Here's an example using @Duck's data (with a different random seed) to show that it is possible to plot the three time series without faking the years, just plugging the dates and values into ggplot and using a couple of lubridate functions:

ggplot(df, aes(x = lubridate::yday(Date), 
               y = value, 
               color = factor(lubridate::year(Date)))) +
geom_line() +
scale_x_continuous(
  breaks = lubridate::yday(seq(as.Date("2017-01-01"), 
                               by = "1 month", length.out = 12)), 
  labels = month.abb) +
  labs(x = "Date", colour = "Year")

Answer 3

in a regular plot it would look something like this:

#x11() # opens a graphics window
plot(mtcars$drat, type="l")
lines(mtcars$wt, col=2) # permits extra lines to be added to an already existing plot
points(mtcars$gear, col=3) # permits points added to an already existing plot

although if you have a dedicated time variable it would change like so:

x11()
plot(mtcars$date, mtcars$drat, type="l")
lines(mtcars$date, mtcars$wt, col=2)
lines(mtcars$date, mtcars$gear, col=3)

as mtcars doesn't have a timeline, this is not a valid example but the idea behind it works