Select 日期基于 dataframe 中的某些标准

Question

I have a dataframe comprised by 4 columns, with temperature data from 3 different locations, from different periods rbind -ed in a single data frame. I want to select temperatures which are in common dates/time(hours) from the 3 stations.

Below I provide a reproducible example:

a1 <- seq.POSIXt(as.POSIXct("1995-01-01"), as.POSIXct("2007-04-01"), by = "120 min")
a2 <- seq.POSIXt(as.POSIXct("1998-04-19"), as.POSIXct("2004-03-20"), by = "60 min")
a3 <- seq.POSIXt(as.POSIXct("1991-01-01"), as.POSIXct("2001-04-01"), by = "180 min")


t1 <- runif(length(a1), min = -5, max = 45)
t2 <- runif(length(a2), min = -5, max = 45)
t3 <- runif(length(a3), min = -5, max = 45)


station1 <- data.frame(date = a1, temp = t1, ID = "station1")
station2 <- data.frame(date = a2, temp = t2, ID = "station2")
station3 <- data.frame(date = a3, temp = t3, ID = "station3")

all_stat <- rbind(station1,station2,station3)


all_stat <- all_stat %>%
  mutate(time = hms::as_hms(date),
         date = as_date(date)) %>%
  relocate(date, time)

Idealy I would like to have the four columns data frame (date/time/temp/ID) with only the common dates/hours of temp data, among these 3 stations. I tried multiple things with dplyr as well as subset but nothing worked.

Answer 1

Combine date and time to create datetime column. split the datetime variable for each ID and find the common ones using Reduce and use it to subset the dataframes to keep only the common date and times between all the ID 's.

all_stat$datetime <- paste(all_stat$date, all_stat$time)
result <- subset(all_stat, datetime %in% 
                    Reduce(intersect, split(all_stat$datetime, all_stat$ID)))

Answer 2

We can do this in tidyverse .

library(dplyr)   
library(stringr)
all_stat %>%       
    group_by(datetime = str_c(date, time)) %>%
    filter(n_distinct(ID) == n_distinct(all_stat$ID))

---

Or if we want to make this faster, use data.table

library(data.table)
setDT(all_stat)[, datetime := paste(date, time)]
sub_stat <- all_stat[all_stat[, .I[uniqueN(ID) == uniqueN(all_stat$ID)],
             by = datetime]$V1]