With R 1.8.1 running in Fedora Core 1 Linux, I am having some trouble recoding and ploting some factor variables. First, can I give you some example data? Here is a column of names for age groups: agegroups <- c( "15-19", "20-24", "25-29","30-34", "35-39", "40-44","45-49","50-54","55-59","60-64", "65-69", "70-74", "75-79", "80-84", "OVER") Here is an index for driver's license ownership in each age group: fracld 0.4914204 0.9752746 1.0864465 1.0555984 1.0631969 1.0738725 1.0971969 1.0657212 1.0217373 0.9761226 0.9043233 0.9045744 0.8573243 0.7889182 0.5217992 I want to take several similar columns of numbers and put them into a single line plot, one line for each column. I can get a graph with the inside part that looks roughly like I want if I just do: plot(fracld,type="l") The horizontal axis, of course, is just the sequence from 1:15, not the age labels. That's no good. But, If I try plot(as.factor(agegroup), fracld, type="l") the plot does not have the line I want, but rather only flat "steps" showing the values. It does have a nice looking horizontal axis, though, showing the age groups. So I think to myself "I'll outsmart them by adding the lines after creating the plot", but if I do this plot(agegroup,fracld,type="n") The step markers still appear. So if I want the tick marks and value lables on the horzontal axis, there is apparently no way to plot lines? What to do?
Trouble plotting with factor
4 messages · Paul Johnson, Marc Schwartz, Martin Maechler
On Fri, 2004-01-30 at 15:04, Paul Johnson wrote:
With R 1.8.1 running in Fedora Core 1 Linux, I am having some trouble recoding and ploting some factor variables. First, can I give you some example data? Here is a column of names for age groups: agegroups <- c( "15-19", "20-24", "25-29","30-34", "35-39", "40-44","45-49","50-54","55-59","60-64", "65-69", "70-74", "75-79", "80-84", "OVER") Here is an index for driver's license ownership in each age group: fracld 0.4914204 0.9752746 1.0864465 1.0555984 1.0631969 1.0738725 1.0971969 1.0657212 1.0217373 0.9761226 0.9043233 0.9045744 0.8573243 0.7889182 0.5217992 I want to take several similar columns of numbers and put them into a single line plot, one line for each column. I can get a graph with the inside part that looks roughly like I want if I just do: plot(fracld,type="l") The horizontal axis, of course, is just the sequence from 1:15, not the age labels. That's no good. But, If I try plot(as.factor(agegroup), fracld, type="l") the plot does not have the line I want, but rather only flat "steps" showing the values. It does have a nice looking horizontal axis, though, showing the age groups. So I think to myself "I'll outsmart them by adding the lines after creating the plot", but if I do this plot(agegroup,fracld,type="n") The step markers still appear. So if I want the tick marks and value lables on the horzontal axis, there is apparently no way to plot lines? What to do?
Paul,
I did not see any responses come through yet on this, so I don't know if
you got anything offlist.
I do not know how your data is structured, but one approach, if your
data is in a matrix, with the numbers being the columns and the
agegroups being the rownames, is the following:
# Create the agegroups
agegroups <- c("15-19", "20-24", "25-29","30-34", "35-39",
"40-44","45-49","50-54","55-59","60-64",
"65-69", "70-74", "75-79", "80-84", "OVER")
# Create the first column
fracld <- c(0.4914204, 0.9752746, 1.0864465, 1.0555984, 1.0631969,
1.0738725, 1.0971969, 1.0657212, 1.0217373, 0.9761226,
0.9043233, 0.9045744, 0.8573243, 0.7889182, 0.5217992)
# Now create two additional columns for the example
# 'fracld2' won't make sense "real world" since many vals
# will be > 1.0
fracld1 <- fracld - 0.25
fracld2 <- fracld + 0.25
# Create the matrix
df <- cbind(fracld, fracld1, fracld2)
# Set the rownames
rownames(df) <- agegroups
df
fracld fracld1 fracld2
15-19 0.4914204 0.2414204 0.7414204
20-24 0.9752746 0.7252746 1.2252746
25-29 1.0864465 0.8364465 1.3364465
30-34 1.0555984 0.8055984 1.3055984
35-39 1.0631969 0.8131969 1.3131969
40-44 1.0738725 0.8238725 1.3238725
45-49 1.0971969 0.8471969 1.3471969
50-54 1.0657212 0.8157212 1.3157212
55-59 1.0217373 0.7717373 1.2717373
60-64 0.9761226 0.7261226 1.2261226
65-69 0.9043233 0.6543233 1.1543233
70-74 0.9045744 0.6545744 1.1545744
75-79 0.8573243 0.6073243 1.1073243
80-84 0.7889182 0.5389182 1.0389182
OVER 0.5217992 0.2717992 0.7717992
# Now use matplot() to plot each column
# Do not plot the axes
matplot(df, type = "l", axes = FALSE,
xlab = "Age Group", ylab = "Proportion DL Ownership")
# Create the X axis, specifying 15 tick marks
# and using rownames(df) as the labels
axis(1, at = 1:nrow(df), labels = rownames(df))
# Now draw the Y axis with defaults
axis(2)
# Put a box around the whole thing
box()
BTW, this is also using FC1 and R 1.8.1 Patched.
HTH,
Marc Schwartz
"Marc" == Marc Schwartz <MSchwartz at medanalytics.com>
on Fri, 30 Jan 2004 20:19:52 -0600 writes:
<<excellent nicely told help text for Paul Johnson>>
I want to comment on the following because it's
"not quite optimal", and still is recommended on-and-on ....
Marc> # Now use matplot() to plot each column
Marc> # Do not plot the axes
Marc> matplot(df, type = "l", axes = FALSE,
Marc> xlab = "Age Group", ylab = "Proportion DL Ownership")
Marc> # Create the X axis, specifying 15 tick marks
Marc> # and using rownames(df) as the labels
Marc> axis(1, at = 1:nrow(df), labels = rownames(df))
Marc> # Now draw the Y axis with defaults
Marc> axis(2)
Marc> # Put a box around the whole thing
Marc> box()
More elegant is not to set axes = FALSE (and having to add
axis(2) and box() later) but to use xaxt = 'n' {only
suppressing x-axis}, i.e., instead of the above 4 statements,
only two :
matplot(df, type = "l", xaxt = "n", # do not plot the 'x' axis
xlab = "Age Group", ylab = "Proportion DL Ownership")
# Create the X axis, specifying 15 tick marks
# and using rownames(df) as the labels
axis(1, at = 1:nrow(df), labels = rownames(df))
Martin
On Sat, 2004-01-31 at 07:34, Martin Maechler wrote:
"Marc" == Marc Schwartz <MSchwartz at medanalytics.com>
on Fri, 30 Jan 2004 20:19:52 -0600 writes:
<<excellent nicely told help text for Paul Johnson>>
Thanks Martin.
I want to comment on the following because it's
"not quite optimal", and still is recommended on-and-on ....
Marc> # Now use matplot() to plot each column
Marc> # Do not plot the axes
Marc> matplot(df, type = "l", axes = FALSE,
Marc> xlab = "Age Group", ylab = "Proportion DL Ownership")
Marc> # Create the X axis, specifying 15 tick marks
Marc> # and using rownames(df) as the labels
Marc> axis(1, at = 1:nrow(df), labels = rownames(df))
Marc> # Now draw the Y axis with defaults
Marc> axis(2)
Marc> # Put a box around the whole thing
Marc> box()
More elegant is not to set axes = FALSE (and having to add
axis(2) and box() later) but to use xaxt = 'n' {only
suppressing x-axis}, i.e., instead of the above 4 statements,
only two :
matplot(df, type = "l", xaxt = "n", # do not plot the 'x' axis
xlab = "Age Group", ylab = "Proportion DL Ownership")
# Create the X axis, specifying 15 tick marks
# and using rownames(df) as the labels
axis(1, at = 1:nrow(df), labels = rownames(df))
Right. I had actually started to use that approach (which I have used in the past for plot(), etc.). I must have had a transient loss of blood flow to the brain, as I for some reason (uncommented in my code), I decided to use 'axes = FALSE'. Either that, or I was asleep at the keyboard and my fingers were on autopilot.... :-) Thanks for pointing that out Martin. Best regards, Marc