Incorrect x axis label positioning

Question

I have a question about axis label positioning that I haven't been able to resolve.

I am using geom_tile plotting for a time series data. I am basically plotting the zone that the animal was in relative to the time course of the experiment. Below is an example df:

   time speed zone   
  <dbl> <dbl> <chr>  
1 0        NA eating 
2 0.3      NA eating 
3 0.334    NA eating 
4 0.368    NA nest
5 0.401    NA nest
6 0.434    NA nest

The plotting works just fine, however, when I get the graph the x axis begins and finishes outside of the actual experiment time. All experiments start precisely at 0 and finish precisely at 1800. However, the graph looks like this:

Can anyone help me with that? If I set the xlim (0, 1800) it works fine, but then the breaks are at 500, 1000 and 1500, whereas I would like them at 600, 1200 and 1800.

Here is the code I'm using to plot:

ggplot(df, aes(x = time, y = animal, fill=zone, width = 200)) +
  geom_tile() +
  facet_wrap(.~week, ncol=1) +
  scale_fill_brewer(palette = "BrBG") +
  scale_x_time(expand = c(0, 0)) +
  scale_y_discrete(expand = c(0, 0)) 

Thank you in advance!

Answer 1

geom_tile isn't the right tool here. It will give you a fixed width centred on the time, but what you need is a variable width rectangle with a left edge at time and a right edge at the next time, so you should perhaps use geom_rect:

ggplot(df, aes(x = time, y = animal, fill = zone)) +
  geom_rect(aes(xmin = time, xmax = c(time[-1], tail(time, 1)), 
                ymin = as.numeric(factor(animal)) - 0.5, 
                ymax = as.numeric(factor(animal)) + 0.5)) +
  facet_wrap(.~week, ncol=1) +
  scale_fill_brewer(palette = "BrBG", drop = FALSE) +
  scale_x_time(expand = c(0, 0), breaks = 60 * c(0, 360, 720, 1080), 
               limits = c(0, 68000)) +
  scale_y_discrete(expand = c(0, 0)) 

Note you didn't have enough sample data to demonstrate this properly, so I created some:

set.seed(1)
df <- data.frame(time = cumsum(runif(50, 0, 2500)),
                 zone = factor(sample(c('eating', 'nest', 'foraging'), 50, TRUE),
                               c('eating', 'nest', 'mating', 'foraging')),
                 week = "Week 1", animal = "Animal 1")

Answer 2

geom_tile creates a rectangle in which x and y represent the center of the tile. This results into an x axis showing data before 0 and after 1800. Use geom_rect instead:

library(tidyverse)

df <- tibble(
  time = c(0, 200,1500,1800),
  zone = c("eating", "eating", "nest", "eating"),
  week = 1,
  animal = "ctrl4"
)

ggplot(df) +
  geom_rect(aes(xmin = time, xmax = lead(time), ymin = 0, ymax= 1, fill = zone)) +
  facet_wrap(.~week, ncol=1) +
  scale_y_discrete(expand = c(0, 0)) +
  scale_x_continuous(expand = c(0,0))
#> Warning: Removed 1 rows containing missing values (geom_rect).

^{Created on 2022-05-20 by the reprex package (v2.0.0)}

Answer 3

If the client-side is using the Android framework, so it won't be affected by porting to AIDL. Because you're porting the HAL implemantation.