Regression line lost after factor conversion - r

Question

In the following plot, time is on the x-axis but tick marks do not show for every year:
ggplot(mm, aes(x = time, y = value)) +
geom_point(aes(color = variable)) +
geom_line(stat = "smooth", method = "lm", alpha = 0.5) +
facet_grid(variable ~ ., scales = "free_y") +
theme(legend.position="none") +
coord_fixed(ratio = 10)
In order to fix this, I have converted the time variable to a factor, which works but then the linear regression disappears:
ggplot(mm, aes(x = factor(time), y = value)) +
geom_point(aes(color = variable)) +
geom_line(stat = "smooth", method = "lm", alpha = 0.5) +
facet_grid(variable ~ ., scales = "free_y") +
theme(legend.position = "none") +
coord_fixed(ratio = 10)
Is there a workaround for this with geom_line?

Answer 1

I think that scale_x_date is what you are looking for.
First, some reproducible data:
df <-
data.frame(
y = 99:117
, x = seq(as.Date("1999-01-01")
, as.Date("2017-01-01")
, "year")
)
Then, this is the way you can set to some "pretty" break points while still getting a tick at each year. If you want every year labelled, then use date_breaks = "1 year" instead of the breaks and date_minor_breaks arguments I have now
ggplot(df, aes(x = x, y = y) ) +
geom_smooth(method = "lm") +
geom_point() +
scale_x_date(breaks = pretty(df$x)
, date_minor_breaks = "1 year"
, date_labels = "%Y")
gives
Or, if your years are just numeric (and not dates), you can use scale_x_continuous for a similar effect:
df <-
data.frame(
y = 99:117
, x = 1999:2017
)
ggplot(df, aes(x = x, y = y) ) +
geom_smooth(method = "lm") +
geom_point() +
scale_x_continuous(breaks = pretty(df$x)
, minor_breaks = unique(df$x)) +
theme_gray()
Gives a plot that is indistinguishable from above.