R和dplyr：将格式变量'dttm'转换为'POSIXCT'

Question

使用apache-spark，这些变量具有一种奇怪的格式，称为dttm，显示如下：

 tpep_pickup_datetime  tpep_dropoff_datetime 
               <dttm>                 <dttm>               
  2015-01-15 18:05:39    2015-01-15 18:23:42                    
  2015-01-10 19:33:38    2015-01-10 19:53:28              
  2015-01-10 19:33:38    2015-01-10 19:43:41              
  2015-01-10 19:33:39    2015-01-10 19:35:31               

我想以秒为单位计算tpep_pickup_datetime和tpep_dropoff_datetime之间的时间差。

但是使用lubridate包无效。如何将这些变量转换为POSIXCT格式使用dplyr？

当我使用以下代码时：

my_df %>% mutate(diff_time = difftime(tpep_dropoff_datetime,tpep_pickup_datetime,units = "secs"))

我收到此错误：

org.apache.spark.sql.catalyst.parser.ParseException: extraneous input 'AS' expecting {')', ','}(line 1, pos 121)

Answer 1

尝试一下：

  all_data <- all_data %>%
      mutate(new_pickup = as.POSIXct(tpep_pickup_datetime)) %>%
      mutate(day_pickup = as.Date(new_pickup)) %>%
      mutate(time_pickup = paste(hour(new_pickup), minute(new_pickup),second(new_pickup),sep="-")) %>%
      mutate(new_dropoff = as.POSIXct(tpep_dropoff_datetime)) %>%
      mutate(day_dropoff = as.Date(new_dropoff)) %>%
      mutate(time_dropoff = paste(hour(new_dropoff), minute(new_dropoff),second(new_dropoff),sep="-")) %>%
      mutate(trip_duration = ((hour(new_dropoff) - hour(new_pickup))*3600 + (minute(new_dropoff) - minute(new_pickup))*60 + (second(new_dropoff) - second(new_pickup))))