如何通过R中的ID计算日期之间的平均差异

Question

I have the data set like below and i want to calculate the average time difference for each unique id

data:
   membership_id created_date 
1       12000000 2015-01-20   
2       12000001 2012-11-19   
3       12000001 2013-10-07   
4       12000001 2014-03-06   
5       12000001 2015-01-14   
6       12000003 2013-02-08   
7       12000003 2014-03-06
8       12000000 2014-02-05
9       12000000 2012-01-06

From the above data set i want to calculate the average time difference between dates for each unique id

TRIED:

 library(plyr)
 data =data[order(data$membership_id,data$created_date),]
 result = ddply(data,.(membership_id),summarize, avg =  as.numeric(mean(diff(created_date))))

The above code is working fine when i am applying on the small data,but my data set is 5 million rows and it is taking lot of time and still it is running from last 6 hours

Expected output:

  membership_id  avg_time_diff
 1 12000000       76 days
 2 12000001       56 days
 3 12000003       54 days

Answer 1

Coming from plyr , you can probably transition very easily to dplyr . It won't be quite as fast as data table, but it will be much faster than ddply .

dat %>% group_by(membership_id) %>%
    arrange(created_date) %>%
    summarize(avg = as.numeric(mean(diff(created_date))))
# Source: local data frame [3 x 2]
#
#   membership_id   avg
#           (int) (dbl)
# 1      12000000   555
# 2      12000001   262
# 3      12000003   391

Without any more real effort, you can speed things up even more by converting to a data.table object but still use the dplyr commands. Pure data.table will still be even faster.

(Using this data)

dat = structure(list(membership_id = c(12000000L, 12000001L, 12000001L, 
12000001L, 12000001L, 12000003L, 12000003L, 12000000L, 12000000L
), created_date = structure(c(16455, 15663, 15985, 16135, 16449, 
15744, 16135, 16106, 15345), class = "Date")), .Names = c("membership_id", 
"created_date"), row.names = c("1", "2", "3", "4", "5", "6", 
"7", "8", "9"), class = "data.frame")