Beginning R date plots and analysis with Matrix object?

Question

I want to plot the data as a time series (quarterly) but the plot keeps treating the dates as Data not labels or time?

Trying to keep my initial exercise very simple, (have more complex stuff to get on with, it would be a snap with Excel or even GDocs!) I am reading a very small table from csv (using read.csv) as

This is the display for Corl (my data in r)

    Date Survey Actual
1 2011-06-30     60     NA
2 2011-03-31     55     50
3 2010-12-31     48     44
4 2010-09-30     48     36
5 2010-06-30     56     75
6 2010-03-31     57     41

I tried to convert to date using Corl$Date <- as.Date(Corl$Date) but no difference.

I have used both plot(Corl) and plot.ts(Corl)

I'll then want to run correlate and auto correlate but I seem to be missing a basic concept for the data structure?

I also tried inverting the columns and rows,

      V1         V2         V3         V4         V5         V6
1   Date 2010-03-31 2010-06-30 2010-09-30 2010-12-31 2011-03-31
2 Survey         57         56         48         48         55
3 Actual         41         75         36         44         50
      V7
1 2011-06-30
2         60
3           

dput(Corl)
structure(list(Date = structure(c(15155, 15064, 14974, 14882, 
14790, 14699), class = "Date"), Survey = c(60L, 55L, 48L, 48L, 
56L, 57L), Actual = c(NA, 50L, 44L, 36L, 75L, 41L)), .Names = c("Date", 
"Survey", "Actual"), row.names = c(NA, -6L), class = "data.frame")

Plot as follows: BadChart Plot

It would help if you could post the output of `dput(Corl)` so we can see how the Date information is currently stored in Corl (as a factor, character, etc). If it is nothing unusual, then the answer will be similar to http://stackoverflow.com/questions/6242955/converting-year-and-month-to-a-date-in-r — bill_080, Jun 05 '11 at 18:56
Looked at the other answer and still don't get it? The output you requested has been added and the graph is my current output of plot(Corl) — dartdog, Jun 05 '11 at 19:11

bill_080 · Accepted Answer · 2011-06-05T20:19:01.150

From the output of dput(Corl) we don't have to guess at how the Date values are stored in Corl. As you can see below, the Dates are stored as class = "Date". That makes this fairly easy.

library(zoo)

Corl <- structure(list(Date = structure(c(15155, 15064, 14974, 14882,
  14790, 14699), class = "Date"), Survey = c(60L, 55L, 48L, 48L,
    56L, 57L), Actual = c(NA, 50L, 44L, 36L, 75L, 41L)), .Names = c("Date",
      "Survey", "Actual"), row.names = c(NA, -6L), class = "data.frame")

Corl.zoo <- read.zoo(Corl, FUN=as.yearqtr)
Corl.zoo

        Survey Actual
2010 Q1     57     41
2010 Q2     56     75
2010 Q3     48     36
2010 Q4     48     44
2011 Q1     55     50
2011 Q2     60     NA

plot(Corl.zoo)

enter image description here

Edit 1 ======================================================

Before you ask, here's a way to add vertical grid lines and more labels across the x axis. All I did was modify the example code at the bottom of the following link:

http://rss.acs.unt.edu/Rdoc/library/zoo/html/plot.zoo.html

my.panel <- function(...) {
   lines(...)
   #This line adds the vertical grid lines
   abline(v=time(Corl.zoo), col="lightgray", lty=3)
   panel.number <- parent.frame()$panel.number
   # if bottom panel
   if (!length(panel.number) || panel.number == NCOL(Corl.zoo)) {
      # next line only if non-labelled ticks wanted for each point
      axis(1, at = time(Corl.zoo), lab = FALSE)
      labcou <- 1 #Put a label, counting between labels
      ix <- seq(1, length(Corl.zoo), labcou)
      labs <- format(time(Corl.zoo), "%Y\nQ%q\n ")
      axis(1, at = time(Corl.zoo)[ix], lab = labs[ix], tcl = -0.5, padj=0.6, cex.axis = 1)
   }
}

plot(Corl.zoo, panel = my.panel, xaxt = "n", main="My Title", xlab="")

enter image description here

Fantastic,, what the heck is zoo?? I'll look it up but thought I'd mention it here from the other answer you linked to it seems that it is some sort of magic function? and How can I plot both lines on the same graph? — dartdog, Jun 05 '11 at 19:28
@dartdog, Zoo is at http://cran.r-project.org/web/packages/zoo/index.html It is worthwhile to read the vignettes. Zoo is well written code with a lot of power. — bill_080, Jun 05 '11 at 19:38
Thank you so much,I was just doing the research! BTW to get both lines in the same panel >> plot(Corl.zoo, plot.type = "single") In case someone else looks at this. — dartdog, Jun 05 '11 at 21:01

Beginning R date plots and analysis with Matrix object?

1 Answers1