x-axis labeling with 'non-date' x-values

Question

I am using R to produce a line graph and have two vectors, one for the x value and another for the y value. More concrete it's accumulated performance over time. I can plot the line using my data (data$AccBH). However, when I try to put the data for y in the code like this:

plot(data$Date, data$AccBH, type='l')

It doesn't work. In the documentation I read this "plot(x, y, ...)" so I thought I could just need to put the vector for y and x in their respective positions. My data for x are 'nondate' dates for example 'Jan-00'.

Basically, all I want is to have some dates from my y data show up on the x axis in stead of the 50, 100, 150.. in the graph below. All I found so far was working with 'real dates' in the x axis.

I just realised I can't post images yet. So I should describe my graph first.

It would be great if somebody could help me out. Also, this is my first question I posted, so if you think I should change the way to post a question, please let me know. Thank you!

Answer 1

You can't expect R and plot() to make up numeric values from arbitrary character strings that you as a human understand but might as well be gibberish to a computer!

You need to inform R that the vector Date in your data frame represents a date - in R parlance an object of class "Date". Now as you don't have days there, we have to make them up - the first of each month will suffice. Consider this simple example data set

df <- data.frame(Date = paste(month.abb, "13", sep = "-"),
                 AccBH = rnorm(12))

I inform R that Date is a "Date" object through as.Date(). Notice here I am pasting on 01- to each of the month-year combinations already in df. An appropriate format specifier is also provided to tell R how to "read" the elements of the date (read ?strftime for those details)

df <- transform(df,
                Date = as.Date(paste("01", as.character(df$Date), sep = "-"),
                               format = "%d-%b-%y"))

Once we have the data in the correct format, as shown here by the str() function

> str(df)
'data.frame':   12 obs. of  2 variables:
 $ Date : Date, format: "2013-01-01" "2013-02-01" ...
 $ AccBH: num  0.494 -0.759 -2.204 -2.004 2.808 ...

plot() works appropriately:

plot(AccBH ~ Date, data = df, type = "l")

producing something like (you'll have different data):

If you want different labelling, then you'll need to specify it yourself, e.g.

plot(AccBH ~ Date, data = df, type = "l", xaxt = "n")
with(df, axis.Date(side = 1, x = Date, format = "%b-%y"))

See ?axis.Date for more details.

Note there is the zoo package which has a data type specifically for this year-month, the "yearmon" class. You can use this without fiddling with your original data

df2 <- data.frame(Date = paste(month.abb, "13", sep = "-"),
                  AccBH = rnorm(12))

through the use of the as.yearmon() function

require("zoo")
df2 <- transform(df2, Date = as.yearmon(Date, format = "%b-%y"))

Then create a "zoo" object using the variables thus created

zobj <- with(df2, zoo(AccBH, order.by = Date))

This gives

> zobj
    Jan 2013     Feb 2013     Mar 2013     Apr 2013     May 2013     Jun 2013 
-0.607986011  1.741046878 -0.603960213 -1.319397405 -0.355051912  0.296862314 
    Jul 2013     Aug 2013     Sep 2013     Oct 2013     Nov 2013     Dec 2013 
 0.235978782 -0.308123299 -0.200230379 -0.004428621 -0.858600654 -0.241808594 

Which can then be plotted

plot(zobj)

but note it has it's own way of labelling the axes.