R strptime 没有管道进入数据帧

Question

library(dplyr)

names <- c('a','b','c')
dates1 <- c('2020-08-14','2020-08-15','2020-08-16')
dates2 <- c('2019-08-14','2019-08-15','2019-08-16')

df <- data.frame(names, dates1, dates2)
print(colnames(df))

timestamps <- df %>% select(dates1, dates2) %>%
  strptime('%Y-%m-%d')
print(timestamps)

Why is timestamps a pair of NA s? How do I get it to correctly apply strptime to these datetime strings?

Answer 1

You are applying sptrptime on dataframe instead you should apply it on columns

library(dplyr)
df %>% mutate(across(starts_with('date'), strptime, '%Y-%m-%d'))

#  names     dates1     dates2
#1     a 2020-08-14 2019-08-14
#2     b 2020-08-15 2019-08-15
#3     c 2020-08-16 2019-08-16

Since you have only date information in the columns you can use as.Date :

df %>% mutate(across(starts_with('date'), as.Date))

Answer 2

It is a two column data.frame, and we could use as.POSIXct if the intention is to convert to DateTime class by looping over the columns with across

library(dplyr) # >= 1.00
out <- df %>% 
    select(dates1, dates2)  %>%
    mutate(across(everything(), as.POSIXct))
out
#  dates1     dates2
#1 2020-08-14 2019-08-14
#2 2020-08-15 2019-08-15
#3 2020-08-16 2019-08-16

If we have an earlier version of dplyr , use mutate_at or mutate_all

df %>%
    select(dates1, dates2) %>%
     mutate_all(as.POSIXct)

---

It can be also used with strptime , but beware of the structure and the class as it is POSIXlt

out2 <- df %>% 
    select(dates1, dates2)  %>%
    mutate(across(everything(), strptime, format = '%Y-%m-%d'))

unclass(out2$dates1)
#$sec
#[1] 0 0 0

#$min
#[1] 0 0 0

#$hour
#[1] 0 0 0

#$mday
#[1] 14 15 16
#...

---

Regarding the OP's original question about piping, it does on a single column or vector

df %>%
    pull(dates1) %>% 
    strptime(format = '%Y-%m-%d')
#[1] "2020-08-14 CDT" "2020-08-15 CDT" "2020-08-16 CDT"

because the documentation for ?strptime says the input should be

x - An object to be converted: a character vector for strptime, an object which can be converted to "POSIXlt" for strftime.

---

If we don't want to select , use the select modifiers

df %>%       
    mutate(across(-1, as.POSIXct))
# names     dates1     dates2
#1     a 2020-08-14 2019-08-14
#2     b 2020-08-15 2019-08-15
#3     c 2020-08-16 2019-08-16

---

Or if the intention is to convert to Date class, just use as.Date

df %>% 
    select(dates1, dates2)  %>%
     mutate(across(everything(), as.Date))

---

NOTE: strptime returns a list and is not recommended

---

Also, another option is base R

df[-1] <- lapply(df[-1], strptime, format = '%Y-%m-%d')