如何在R中的个人中为不同的观察日期添加独特的场合？

Question

I'm trying to figure out how in my data set to add a column including a count of unique events based on date within each patient. Here is a part of my data set:

    trialno     event          date          time
3   11301   pm_intake   2010-11-24  19:00
4   11301   am_intake   2010-11-25  07:00
5   11301   pk1         2010-11-25  10:30
6   11301   pm_intake   2010-12-22  19:00
7   11301   am_intake   2010-12-23  07:00
8   11301   pk1         2010-12-23  09:54
9   11301   pk2         2010-12-23  13:07
10  11301   pm_intake   2011-02-02  19:00
11  11301   am_intake   2011-02-03  07:00
12  11301   pk1         2011-02-03  11:30

Basically each date within each patient would indicate a new occasion. If patient has just drug administration - it's one occasion but if patient had drug administration and two measurements on the same day, they all count as the same occasion. The data set does not have a regular patters (each patient has a different number of events on each date and events in total). What I'm trying to achieve is:

    trialno     event          date          time   OCC
3   11301   pm_intake   2010-11-24  19:00    1
4   11301   am_intake   2010-11-25  07:00    2
5   11301   pk1         2010-11-25  10:30    2
6   11301   pm_intake   2010-12-22  19:00    3
7   11301   am_intake   2010-12-23  07:00    4
8   11301   pk1         2010-12-23  09:54    4
9   11301   pk2         2010-12-23  13:07    4
10  11301   pm_intake   2011-02-02  19:00    5
11  11301   am_intake   2011-02-03  07:00    6
12  11301   pk1         2011-02-03  11:30    6

I think I should apply some kind of a loop to identify within each patient unique dates and count them but I'm not sure how to write it, so I tried using apply function.

I thought about splitting the whole data set into individual patients first using split function:

splitData<- split(data, data$trialno)

And applying lapply and transform to add a new column OCC (occasion) but I don't know how to count those as integers...

I was thinking:

splitData<- lapply(splitData, function(df) {
       transform(df, OCC= ???????????????   )}

do.call ("rbind", splitData)

I know how to do it in Excell:

=IF(D5=D4, E4,E4+1)

(if the cell value in neighbouring cell is same as in the cell above, then value in my cell is same as in one above, else it's one greater)-this way first cell in E column has to be 1 and the others are integers of new date events.

I tried looking for similar questions on stack overflow but without any luck.

Help much appreciated!

Answer 1

If I understand correctly, what you want is for OCC to indicate unique dates for each trial number, but you want to restart OCC at 1 for each new trial number. This can be accomplished most easily using the data.table package.

First I'll generate some data with multiple trial numbers:

> dt0
   trialno     event       date  time
1    11301 pm_intake 2010-11-24 19:00
2    11301 am_intake 2010-11-25 07:00
3    11301       pk1 2010-11-25 10:30
4    11301 pm_intake 2010-12-22 19:00
5    11301 am_intake 2010-12-23 07:00
6    11301       pk1 2010-12-23 09:54
7    11301       pk2 2010-12-23 13:07
8    11301 pm_intake 2011-02-02 19:00
9    11301 am_intake 2011-02-03 07:00
10   11301       pk1 2011-02-03 11:30
11   11302       pk1 2011-02-03 11:30
12   11302       pk1 2011-02-03 11:40

The OCC column can be added like this:

> require(data.table)
> dt<-data.table(dt0)
> dt[,OCC:=match(date,unique(date)),by=trialno]
> dt
    trialno     event       date  time OCC
 1:   11301 pm_intake 2010-11-24 19:00   1
 2:   11301 am_intake 2010-11-25 07:00   2
 3:   11301       pk1 2010-11-25 10:30   2
 4:   11301 pm_intake 2010-12-22 19:00   3
 5:   11301 am_intake 2010-12-23 07:00   4
 6:   11301       pk1 2010-12-23 09:54   4
 7:   11301       pk2 2010-12-23 13:07   4
 8:   11301 pm_intake 2011-02-02 19:00   5
 9:   11301 am_intake 2011-02-03 07:00   6
10:   11301       pk1 2011-02-03 11:30   6
11:   11302       pk1 2011-02-03 11:30   1
12:   11302       pk1 2011-02-03 11:40   1