5

I am working with some time-frequency decomposed EEG data and want to produce a spectrogram-like figure using ggplot2. But, I end up with blank spaces between each of my time points.

Data <- read.csv(url("https://www.dropbox.com/s/al3cygigm86mr3s/Test_Spec_Data.csv?dl=0"))

If I create a vanilla geom_raster I get gaps in the x and y data:

ggplot(Data,aes(Times,Frequency)) +
  geom_raster(aes(fill = ERSP))

Default

If I make Frequency a factor, it fills in the y gap; but, the gaps along the x-axis remain:

  ggplot(Data,aes(Times,factor(round(Frequency,digits=1)))) +
  geom_raster(aes(fill = ERSP))

With y as factor

I can eliminate the gaps by making Times a factor.

enter image description here

But, managing scale_x_discrete with this many data points is cumbersome (note the x-axis labels). Also, these time data are continuous and not really factor-like.

geom_raster doesn't have a width argument like geom_bar and I can't see anything similar in the geom_raster documentation.

Is there a way to keep Times as continuous but remove the gaps between observations?

JLC
  • 661
  • 7
  • 16

1 Answers1

4

There are gaps because there are not enough data (or more precisely, they are not evenly spaced).

Your "factor" transformation remove the gap because it remove the X or Y axis parts where data are missing. See in Y axis: ticks are evenly spaces but values are not (8.5-8.1=0.4, while 11.3-10.7=0.6, where you have you biggest gap).

I can see two solutions:

  • Interpolate data so your source data are evenly spaced
  • Use geom_tile instead of geom_raster and specify width and height parameters to "expand" your tiles and fill the gaps, as explained in the fourth example of the doc.
Blacksad
  • 1,230
  • 1
  • 14
  • 23