Question

I’ve got some data in the following shape:

UPDATE: My data has an extra variable I’d like to group by. I used ddply with the below solution provided by Richie but did not work.

Country,group, date
US,A,'2011-10-01'
US,B,'2011-10-01'
US,C,'2011-10-01'
MX,D,'2011-10-01'
UK,E,'2011-10-02'
UK,B,'2011-10-02'
UK,A,'2011-10-02'
UK,C,'2011-10-02'

The data frame is already ordered so A came first, B second and so on so forth. What I am trying to create is a rank variable by date like this:

Country,group, date,rank
US,A,'2011-10-01',1
US,B,'2011-10-01',2
US,C,'2011-10-01',3
MX,D,'2011-10-01',1
UK,E,'2011-10-02',1
UK,B,'2011-10-02',2
UK,A,'2011-10-02',3
UK,C,'2011-10-02',4
    ....

Answer 1

First, check that your date really is in a date format (not a factor) using class(your_dataset$date). IF not, use ymd from lubridate to convert it.

Second, use rank to get the rank. (Easier than you think, right!)

your_dataset$rank <- rank(your_dataset$date)

There are a few different methods for breaking ties that you might want to explore.

Upon rereading your question, I see you don’t want to rank the dates, you want a counter within the dates. To do this, first check that your dataset is ordered by date.

o <- with(your_dataset, order(date))
your_dataset <- your_dataset[o, ]

Then call seq_len on each chunk of date.

counts <- as.numeric(table(your_dataset$date))
your_dataset$rank <- unlist(lapply(counts, seq_len))