r - Geom tile collapses along Y-axis when adding coord_equal()

Question

I have a time series data (currently on monthly frequency) from 2014 up to 2020. That means at most 84 monthly values.

For example when plotting (two of categories from the time series data) using geom_tile, I would have a visual below.

ID <- rep(c('A','B','C'),each = 84)
n <- rep(round(runif(84,1,4)), 3)
datetime <- rep(seq(as.POSIXct("2014-01-01"), as.POSIXct("2020-12-01"), by="month"), 3)
df <- tibble(ID,n, datetime)
ggplot(df ,
       aes(x=datetime,y=ID,fill=n))+
  geom_tile() + 
  scale_y_discrete(expand=c(0,0)) + 
  scale_fill_viridis() 

I want the rectangle to be a square, so I add on coord_equal() to my ggplot, but somehow the heatmap collapses along the Y-axis

ggplot(df ,
       aes(x=datetime,y=ID,fill=n))+
  geom_tile() + 
  scale_y_discrete(expand=c(0,0)) + 
  scale_fill_viridis() + coord_equal()

The expectation is something like below. As you can see, the squares are of equal size across.

Answer 1

The reason for why the tile map collapses are well explained in the excellent answer by @eipi10 on the same issue. As the reason is that one unit on the x-axis is by default translated into one unit on the y-axis you could prevent the collapsing of your tile map by setting the conversion ratio according to the ranges of the x and y scale:

ID <- rep(c('A','B','C'),each = 84)
n <- rep(round(runif(84,1,4)), 3)
datetime <- rep(seq(as.POSIXct("2014-01-01"), as.POSIXct("2020-12-01"), by="month"), 3)
df <- data.frame(ID,n, datetime)

library(ggplot2)
library(viridis)
#> Loading required package: viridisLite
ggplot(df ,
       aes(x=datetime,y=ID,fill=n))+
  geom_tile() + 
  scale_y_discrete(expand=c(0,0)) + 
  scale_fill_viridis() +
  coord_equal(ratio = diff(range(as.numeric(datetime))) / length(unique(ID)))