按时间创建图表

Question

I am trying to create a graph in R, that shows how the distance to an object changes by time of day. I have a dataframe in R, that includes datetimes of GPS locations and the distance of those locations to a polygon.

ID                 date1       time      lat       lon    DOP     date      dist         datetime
281 Sunday, October 1, 2017 3:00:15 AM 32.59848 -90.24402 3.8 2017-10-01  57.75744 2017-10-01 03:00:15
281 Sunday, October 1, 2017 2:45:00 AM 32.59851 -90.24391 3.0 2017-10-01  54.07150 2017-10-01 02:45:00
281 Sunday, October 1, 2017 3:15:16 AM 32.59838 -90.24387 4.0 2017-10-01  68.67032 2017-10-01 03:15:16
281 Sunday, October 1, 2017 3:30:12 AM 32.59808 -90.24369 4.2 2017-10-01 103.90344 2017-10-01 03:30:12
281 Sunday, October 1, 2017 3:45:09 AM 32.59815 -90.24354 4.4 2017-10-01 100.03097 2017-10-01 03:45:09
281 Sunday, October 1, 2017 4:00:30 AM 32.59812 -90.24324 2.0 2017-10-01 114.68612 2017-10-01 04:00:30

Using this dataframe, I would like to make a graph showing the how the distance changes over time of day. This way I could see around what time of day my IDs are closer based on distance. This dataset includes 4 months, from October to January. I am imagining a graph like what was created here, http://www.sthda.com/english/articles/32-r-graphics-essentials/128-plot-time-series-data-using-ggplot/ , except it uses time of day instead of year.

Thank you for the help.

Using dput, here are the first 6 rows of my dataset.

structure(list(ID = c("281", "281", "281", "281", "281", "281"
), date1 = c("Sunday, October 1, 2017", "Sunday, October 1, 2017", 
"Sunday, October 1, 2017", "Sunday, October 1, 2017", "Sunday, October 1, 2017", 
"Sunday, October 1, 2017"), time = c("3:00:15 AM", "2:45:00 AM", 
"3:15:16 AM", "3:30:12 AM", "3:45:09 AM", "4:00:30 AM"), lat = c(32.59848, 
32.59851, 32.59838, 32.59808, 32.59815, 32.59812), lon = c(-90.24402, 
-90.24391, -90.24387, -90.24369, -90.24354, -90.24324), DOP = c(3.8, 
3, 4, 4.2, 4.4, 2), date = structure(c(17440, 17440, 17440, 17440, 
17440, 17440), class = "Date"), dist = c(57.7574388004, 54.0715015597, 
68.6703208583, 103.903443285, 100.030967759, 114.686118929)), row.names = c(NA, 
6L), class = "data.frame")

Answer 1

So your time variable is coming in as a character. To my knowledge there's no way to 1:1 extract just the time without making it a date time variable, which would not aggregate it. So you extract the individual pieces, and aggregate in the time_numeric variable.

 library(tidyverse)
 data %>% 
  mutate(hours = as.numeric(str_extract(time, "^\\d+")),
         minutes = as.numeric(str_extract(time, "(?<=\\:)\\d+")),
         seconds = as.numeric(str_extract(time, "(?<=\\:\\d{2}\\:)\\d+")),
         am = ifelse(str_detect(time, "AM"), 0, 12),
         time_numeric = hours + am + minutes/60 + seconds/3600) %>% 
  ggplot(aes(time_numeric, dist)) +
  geom_line() +
  scale_x_continuous(breaks = scales::pretty_breaks(n = 4))

So I added in some logic to round to nearest hour, the round function wrapping time_numeric will round the hour to the nearest whole number. I changed the geom to point so you can see it more easily. If you do geom_line you need something to differentiate the lines otherwise you get a vertical line representing the range of values.

library(tidyverse)
 data %>% 
  mutate(hours = as.numeric(str_extract(time, "^\\d+")),
         minutes = as.numeric(str_extract(time, "(?<=\\:)\\d+")),
         seconds = as.numeric(str_extract(time, "(?<=\\:\\d{2}\\:)\\d+")),
         am = ifelse(str_detect(time, "AM"), 0, 12),
         time_numeric = round(hours + am + minutes/60 + seconds/3600)) %>% 
  ggplot(aes(time_numeric, dist)) +
  geom_point() +
  scale_x_continuous(breaks = scales::pretty_breaks(n = 4))