My data looks like this:
d <- data.frame(X=c('x1','x2','x3','x1','x2','x3','x1','x2','x3'),
Y=c('y1','y1','y1','y2','y2','y2','y3','y3','y3'),
Value=c(1,2,1,3,1,4,3,5,2))
I use the following code to generate a heat map:
ggplot(d,aes(x=X,y=Y,fill=Value)) +
geom_tile() +
scale_fill_gradient2(low='green',mid='white',high='red',midpoint=3) +
theme(axis.text.x = element_text(angle = 90),
axis.text = element_text(size = 10),
panel.background = element_blank()) +
labs(x='',y='')
The graph is:
But, my whole data is around 500*500, so too many ticks on x axis and y axis so that the labels are impossible to recognize clearly. But I have to keep all the values.
So I want to use double x axis and double y axis. For example, the bottom x axis only keeps the lables x1, x3, x5... and the top x axis only keeps the labels x2, x4, x6.... Then I do the same to y axis. Then, the graph looks like:
I know scale_x_continuous and sec.axis might do this. But my x and y are discrete.
Could anyone help me so that the graph looks like the second one?
You are right than you can only use a secondary axis with continuous data. One solution is to convert your x and y values to continuous values for plotting (just use a sequence of integers), but save your discrete x and y values as vectors to use for labelling your axes.
Here I've assigned continuous values in a way that should be scalable to your larger, actual dataset. I've made key dataframes for x and y to match each unique discrete value with an integer, than merged these key dataframes with your original dataframe so each discrete x and y value is assigned a numeric value that you can use for plotting.
#create key dataframes to assign an integer to each x and y value
key.df.x <- data.frame(X = unique(d$X), x.num = (1:length(unique(d$X))))
key.df.y <- data.frame(Y = unique(d$Y), y.num = (1:length(unique(d$Y))))
#merge key dataframes with original data
d <- merge(d, key.df.x, by = "X", all.x = TRUE)
d <- merge(d, key.df.y, by = "Y", all.x = TRUE)
#make label vectors from original variable names
xlabels = unique(d$X)
ylabels = unique(d$Y)
#select odd numbered elements for primary labels, even for secondary labels
xlabels.primary <- xlabels[seq(1, length(xlabels), by = 2)]
xlabels.secondary <- xlabels[seq(2, length(xlabels), by = 2)]
ylabels.primary <- ylabels[seq(1, length(ylabels), by = 2)]
ylabels.secondary <- ylabels[seq(2, length(ylabels), by = 2)]
ggplot(d,aes(x = x.num, y = y.num,fill=Value)) + #plot using continuous data
geom_tile() +
scale_fill_gradient2(low='green',mid='white',high='red',midpoint=3) +
theme(axis.text.x = element_text(angle = 90),
axis.text = element_text(size = 10),
panel.background = element_blank()) +
# set primary axis breaks to odd numbers, label with ylabels.primary
scale_y_continuous(breaks = seq(1, max(d$y.num), by = 2), labels = ylabels.primary,
# set secondary axis breaks to even numbers, label with ylabels.secondary
sec.axis = dup_axis(breaks = seq(2, max(d$y.num), by = 2),
labels = ylabels.secondary)) +
scale_x_continuous(breaks = seq(1, max(d$x.num), by = 2), , labels = xlabels.primary,
sec.axis = dup_axis(breaks = seq(2, max(d$x.num), by = 2),
labels = xlabels.secondary)) +
labs(x='',y='')