计算自上次观察以来的月数

Question

My dataframe look like below:

ID    Date       Value  SinceLastObservation
1     1-1-2010   0      0
1     2-1-2010   0      0
1     3-1-2010   1      0
1     4-1-2010   0      1
1     5-1-2010   1      2
1     7-1-2010   0      2
1     9-1-2010   0      4
2     1-1-2011   1      0
2     2-1-2011   0      1
2     3-1-2011   0      2
2     4-1-2011   1      3
2     6-1-2011   1      2
2     8-1-2011   0      2

I want to add a new column (SinceLastObservation) that counts the number of months from the observations (value "1") in each group (grouped by ID) in R.

This is my solution, but it is super slow for the big data frame that I have.

 DT <- data.table(df)
 DT[, grp := cumsum(Value== "1")- Value, by=list(ID)]
 DT[, minDate := rollback(min(Date), preserve_hms = FALSE),by=list(ID,grp)]
 DT[, Months_since_last_1_30DPD := mondf(minDate,Date),by=list(ID,grp)]

mondf is a function that counts number of months between two dates

The example:

data <- data.frame(ID = c("1", "1", "1","1", "1", "1","1" ,"2", "2","2", "2","2", "2"), 
               Date = c("1-1-2010","2-1-2010","3-1-2010","4-1-2010","5-1-2010","7-1-2010","9-1-2010","1-1-2011","2-1-2011","3-1-2011","4-1-2011","6-1-2011","8-1-2011"), 
               value = c(0,0,1,0,1,0,0,1,0,0,1,1,0))

Thanks

Answer 1

Here is a method using lubridate :

library(tidyverse)
library(lubridate)

data %>% 
  as_tibble %>% 
  mutate(Date = mdy(Date)) %>% 
  group_by(ID) %>% 
  mutate(last_obs = if_else(value==1, Date, NA_Date_),
         last_obs = lag(last_obs)) %>% 
  fill(last_obs) %>% 
  mutate(months_since_last_obs = (last_obs %--% Date)/months(1)) %>% 
  replace_na(list(months_since_last_obs = 0))