Plotting irregular times series with predictable gaps

gaps.R

dat <- structure(list(date = structure(c(14645, 14590, 14577, 14569, 
14289, 14261, 14226, 14219, 14213, 14205, 14295, 14655, 14660, 
14933, 14940, 14941, 14954, 14961, 14974, 14975, 14977, 14980, 
14982, 14983, 15024, 15032, 15034, 15045, 15051, 15052, 15053, 
15297, 15304, 15305, 15318, 15325, 15337, 15338, 15340, 15342, 
15344, 15346, 15347, 15361, 15374, 15375, 15381, 15402, 15409, 
15668, 15675, 15687, 15690, 15703, 15704, 15708, 15709, 15710, 
15711, 15753, 15759, 15767, 15774, 15780, 16039, 16040, 16046, 
16102, 16110, 16115, 16124, 16487, 16491, 16502, 16767, 16768, 
16774, 16781, 16789), class = "Date"), y = c(-2.4215, -2.7139, 
-3.4126, -2.4506, -2.277, 0.1253, -0.9912, 0.0673, -2.5237, -1.7491, 
-5.4418, 2.0619, 6.631, -3.3748, -4.3086, -2.6573, -1.7799, -1.9859, 
1.8146, -0.7082, -0.7949, 0.2595, 10.2503, -2.2439, 1.4828, -1.061, 
-1.8546, -0.5535, 0.5664, -0.5346, 2.2731, 0.5518, -2.5637, 0.512, 
0.9785, 1.6282, 1.261, -1.7903, 0.4518, 1.1885, 0.0195, 5.9399, 
0.6011, -1.558, -0.2032, -0.8846, 0.118, 1.7331, -0.5502, -0.9086, 
-1.3498, -3.0417, -4.1688, -0.8954, 1.0585, 0.9385, 2.5769, 1.2266, 
5.5859, 0.4341, 0.9447, 1.9321, -1.3592, 1.1439, -0.1818, 0.0847, 
5.4002, 3.1371, 4.2006, 1.1091, 10.6764, 5.8082, 3.0979, 4.0353, 
4.0156, 2.4512, 6.5072, 4.4321, 6.29)), class = "data.frame", row.names = c(NA, 
-79L), .Names = c("date", "y"))

#Base graphics
plot(dat)

#ggplot2
library(ggplot2)
ggplot(dat,aes(x = date,y = y)) + 
  geom_point()
  
#Generally the data will only appear for Nov-Apr, never May-Oct.
# Leaving them in often makes the data squashed together and hard to see.
# 1) Leave gaps in or not?
# 2) Are there existing ggplot2 tools for dealing with axes like these?
# 3) Ideas?

My idea was to take the timepoints that you have and space them out using a factor scale, but this really won't work because you have daily data and you would need a vector with every date in between your months. You can do this, but then labeling gets pretty hard too. I know Hadley isn't big on broken axes, but this is an example where it would be useful.

The alternative is faceting, but direct comparisons are harder and your "seasons" aren't Year, they span years. I think this means you need to copy the year (really season) to a column, translate your dates to a single season, then facet by the copied season with a defined date range of a single season. You could color by season to see the data directly on top of each other, or facet_grid( . ~ Season) or facet_grid(Season ~.)

library(tidyverse)
library(lubridate)

dat %>%
mutate(year = year(date), month = month(date), yday = yday(date)) %>%
mutate(
season = if_else(month < 6, year-1, year),
ydayadj = if_else(yday > 275, yday - 365, yday)
) %>%
ggplot(., aes(x = yday, y = y)) +
geom_point() +
facet_grid(.~season)