ggplot2: plot seconds in %H%M%S format in y axi - r

Question

I've a column in seconds that need to plot as "%H%M%S".
I've tried using lubridate pkg, but the column results in:
loadtime_dfs$avgPageLoadTime <- seconds_to_period(loadtime_df$avgPageLoadTime)
Formal class 'Period' [package "lubridate"]
that I can plot but doesn't show any format.
loadtime_df <- structure(list(date = structure(c(17766, 17767, 17768, 17769,
17770, 17771), class = "Date"), pagePath = c("/webapp/wcs/stores/servlet/CategoryDisplay?urlRequestType=Base&catalogId=3074457345616676668&categoryId=3074457345616676994&pageView=grid&urlLangId=-24&beginIndex=0&langId=-24&top_category=3074457345616676981&parent_category_rn=3074457345616720192&storeId=10151",
"/webapp/wcs/stores/servlet/CategoryDisplay?urlRequestType=Base&catalogId=3074457345616676668&categoryId=3074457345616676994&pageView=grid&urlLangId=-24&beginIndex=0&langId=-24&top_category=3074457345616676981&parent_category_rn=3074457345616720192&storeId=10151",
"/webapp/wcs/stores/servlet/CategoryDisplay?urlRequestType=Base&catalogId=3074457345616676668&categoryId=3074457345616676994&pageView=grid&urlLangId=-24&beginIndex=0&langId=-24&top_category=3074457345616676981&parent_category_rn=3074457345616720192&storeId=10151",
"/webapp/wcs/stores/servlet/CategoryDisplay?urlRequestType=Base&catalogId=3074457345616676668&categoryId=3074457345616676994&pageView=grid&urlLangId=-24&beginIndex=0&langId=-24&top_category=3074457345616676981&parent_category_rn=3074457345616720192&storeId=10151",
"/webapp/wcs/stores/servlet/CategoryDisplay?urlRequestType=Base&catalogId=3074457345616676668&categoryId=3074457345616676994&pageView=grid&urlLangId=-24&beginIndex=0&langId=-24&top_category=3074457345616676981&parent_category_rn=3074457345616720192&storeId=10151",
"/webapp/wcs/stores/servlet/CategoryDisplay?urlRequestType=Base&catalogId=3074457345616676668&categoryId=3074457345616676994&pageView=grid&urlLangId=-24&beginIndex=0&langId=-24&top_category=3074457345616676981&parent_category_rn=3074457345616720192&storeId=10151"
), pageviews = c(245L, 225L, 194L, 214L, 214L, 213L), pageLoadTime = c(18965L,
185834L, 31115L, 114561L, 88807L, 0L), avgPageLoadTime = c(6,
27, 16, 138, 144, 0), bouncerate = c(5.63380281690141, 3.48837209302326,
5.40540540540541, 7.01754385964912, 0, 5), mes = c("agosto",
"agosto", "agosto", "agosto", "agosto", "agosto")), .Names = c("date",
"pagePath", "pageviews", "pageLoadTime", "avgPageLoadTime", "bouncerate",
"mes"), row.names = c(NA, -6L), class = c("tbl_df", "tbl", "data.frame"
))
This is what I need to plot:
ggplot(loadtime_df, aes(date,avgPageLoadTime)) +
geom_point() +
geom_smooth()
But with the Y axis with breaks: "00:01:00", "00:02:00", "00:03:00", "00:04:00", "00:05:00".

Answer 1

You are going to have to provide strings for ggplot to assign as labels.
If you read ?scale_y_continuous(labels=...), you'll see that labels= takes either waiver(), character, or a function. If you want to specify the specific locations and representations, then you'll want to specific both breaks= and labels=. However, typically you want ggplot2 to determine where to put the axis labels, so we'll provide a function that takes a value and returns a string.
I'm guessing there's a helper function somewhere to do this, but here's a base-R version. (The origin of the function is unimportant, since we can replace our function with another with likely the same outcome.)
This formatting function cheats a little by temporarily converting the seconds of avgPageLoadTime to POSIXct and then to a string. Doing it this way means it honors options("digits.secs"), if set.
fmt_hms <- function(x, digits.secs=NULL) {
if (!is.null(digits.secs)) {
oopts <- options(digits.secs = digits.secs)
on.exit(options(oopts), add=TRUE)
}
format(as.POSIXct(x, origin="1970-01-01 00:00:00"), format="%H:%M:%OS", tz="UTC")
}
In order to demonstrate this, I'll change one of the values of your data:
loadtime_df$avgPageLoadTime[3] <- loadtime_df$avgPageLoadTime[3] + 0.123456
fmt_hms(loadtime_df$avgPageLoadTime)
# [1] "00:00:06" "00:00:27" "00:00:16" "00:02:18" "00:02:24" "00:00:00"
fmt_hms(loadtime_df$avgPageLoadTime, digits.secs=3)
# [1] "00:00:06.000" "00:00:27.000" "00:00:16.123" "00:02:18.000" "00:02:24.000"
# [6] "00:00:00.000"
So we can just provide this function:
library(ggplot2)
ggplot(loadtime_df, aes(date,avgPageLoadTime)) +
geom_point() +
geom_smooth() +
scale_y_continuous(labels=fmt_hms)

Answer 2

I think you need to convert the date values in seconds to a %H%m%s format and then try plotting. I think you need one of the below approaches -
library(ggplot2)
library(lubridate)
# convert seconds to periods
td <- seconds_to_period(loadtime_df$avgPageLoadTime)
# then apply the required format
avgPageLoadTime_vector <- sprintf('%02d:%02d:%02d', td#hour, minute(td),
second(td))
# plotting using %H%m%s we can use them as y-axis ticks
# this will give you the same plot as above but Y-axis is fuzzy
ggplot(loadtime_df, aes(date,avgPageLoadTime)) +
geom_point() +
geom_smooth() +
scale_y_continuous(breaks = loadtime_df$avgPageLoadTime,
labels = avgPageLoadTime_vector)
# if you just want to plot with points and not use geom_smooth
# convert the column avgPageLoadTime into %H%m%s date-time format
loadtime_df$avgPageLoadTime <- avgPageLoadTime_vector
# this gives you the right Y-axis values but no smoothing
ggplot(loadtime_df, aes(date,avgPageLoadTime)) +
geom_point()
]2