geom_qq() labeling each datapoint with the corresponding 'year' - r

Question

library(fpp3)
library(hrbrthemes)
library(readr)
# load 'Fastenal' dataset and convert to tsibble
Fastenal <- read_csv('http://course1.winona.edu/bdeppa/FIN%20335/Datasets/Fastenal%20Sales%20(2004-2013).csv')
Fastenal %>% mutate(YearMonth = yearmonth('2004-01') + 0:119,
TotSales = Total.Sales/1000000) %>%
as_tsibble(index = YearMonth) -> Fastenal
# Fit AvSalesPD using SNAIVE model
Fastenal %>% model(SNAIVE(AvSalesPD/1000000)) -> fit
# Plot residual QQ plot, label each data points with corresponding 'Year'
fit %>% augment() %>%
ggplot(aes(sample = .resid/1000000)) +
geom_qq(aes(label = year(YearMonth), geom = 'text') + geom_qq_line()+ theme_ipsum()
I'm trying to draw QQ plot, using 'geom = "text"'.
When I run the above codes, I get error message as below;
'Error: geom_text requires the following missing aesthetics: label'
I couldn't find what I did wrong...-_-;

Answer 1

I think the problem is that geom_qq does not know how to use the aesthetics you are providing. By specifing geom = "text" in the function call, it should be drawing text instead of points. The problem is that there is no text supplied that the geom_text can use.
Thus we need to supply the labels. The following code should produce what you asked.
library(fpp3)
library(readr)
# load 'Fastenal' dataset and convert to tsibble
Fastenal <- read_csv('http://course1.winona.edu/bdeppa/FIN%20335/Datasets/Fastenal%20Sales%20(2004-2013).csv')
Fastenal %>% mutate(YearMonth = yearmonth('2004-01') + 0:119,
TotSales = Total.Sales/1000000) %>%
as_tsibble(index = YearMonth) -> Fastenal
# Fit AvSalesPD using SNAIVE model
Fastenal %>% model(SNAIVE(AvSalesPD/1000000)) -> fit
# Plot residual QQ plot, label each data points with corresponding 'Year'
fit <- fit %>% augment()
# here we define the labels to use.
labs <- year(fit$YearMonth)
labs <- labs[!is.na(fit$.resid)]
labs <- labs[order(na.omit(fit$.resid))] #EDIT: important! sort the labels for QQ-plots
ggplot(fit, aes(sample = .resid/1000000)) +
geom_qq(geom = "text", label = labs, angle = 90, size=1.5) +
geom_qq_line()
Note how we are manually creating the labels. We pass to geom_qq the same arguments as you passed, but outside the aes().
Since we are passing this information outside aesthetics, we have to take care that information in labs matches that of the aesthetics used.
Indeed,the points that you are using to draw the QQ plot have some missing entries (12). These are omitted, and as a result you have 108 points (not 120) in your resulting plot. This is why we have to make sure that labs has one entry for each non-missing point in your data (i.e. .resid entries different from NA).
If you do not filter NAs in the construction of labs you end up with an error and no plot.
The above code produces this plot: