I have a dataframe (df) with 3 columns - a stage number, time data, and pressure data. Here is a portion of it:

structure(list(
Stage = c(1, 1, 1, 1, 1, 2, 2, 2), 
Pressure = c(3.24, 12.218, 9.634, 9.027, 9.027, 0, 14.28, 1.737), 
DateTime = structure(c(1624720853, 1624720854, 1624720855, 1624720856, 1624720857, 1624905025, 1624905026, 1624905027), 
tzone = "", class = c("POSIXct", "POSIXt")),
class = "data.frame", 
row.names = c(NA, -8L))

I want to calculate the slope/derivative (change in pressure over change in time) for pressure point for each stage. I have figured out how to calculate the slope, but there are sometimes large gaps in stages and I don't need slope for changes in pressure between stages.

I have code that I believe would work, but because it is looking at the difference in rows, the output will always going to be one less row than than the total number rows within a stage.

df<- df%>% 
  group_by(Stage) %>%
  mutate(dp.dt = diff(Pressure)/as.numeric(diff(DateTime)) )

This is the error, and like I mentioned, I believe it is happening because the code is looking at the difference in rows, which should result in one less row than the true number of rows in a stage:

Error: Problem with `mutate()` column `dp.dt`.
i `dp.dt = diff(Pressure)/as.numeric(diff(DateTime))`.
i `dp.dt` must be size 5 or 1, not 4.
The error occurred in group 1: JobStage = 1.

In the end, I am looking for something like the table below. Is there a way to induce an NA, add a row, or fill the missing row with something so that I get my desired table?

Please let me know if I need to clarify anything. Any help would be appreciated. Thank you.

CodePudding user response：

The diff returns output with length one less than the original data length. Just append NA to the diff and it should work

library(dplyr)
df%>% 
  group_by(Stage) %>%
  mutate(dp.dt = c(diff(Pressure), NA)/as.numeric(c(diff(DateTime), NA)) )

CodePudding user response：

You may use lead/lag to get next and previous value respectively.

library(dplyr)
library(lubridate)

df %>%
  mutate(DateTime = ymd_hms(DateTime)) %>% 
  group_by(Stage) %>%
  mutate(dp.dt = (lead(Pressure) - Pressure)/as.numeric((lead(DateTime) - DateTime))) %>%
  ungroup

#  Stage Pressure DateTime              dp.dt
#  <dbl>    <dbl> <dttm>                <dbl>
#1     1     3.24 2021-06-26 23:20:53   8.98 
#2     1    12.2  2021-06-26 23:20:54  -2.58 
#3     1     9.63 2021-06-26 23:20:55  -0.607
#4     1     9.03 2021-06-26 23:20:56   0    
#5     1     9.03 2021-06-26 23:20:57  NA    
#6     2     0    2021-06-29 02:30:25  14.3  
#7     2    14.3  2021-06-29 02:30:26 -12.5  
#8     2     1.74 2021-06-29 02:30:27  NA