Data table aggregate by week and userid

Question

I'm using R and have a data table DT with 30 million rows:

userid,     date, measurement
   101, 1/1/2011,          13
   101, 2/1/2011,          42
   ...
   333, 1/1/2011,          67
   ...

I'm thinking of aggregating the observations by userid and week.

My current idea is to convert date into an integer, and then divide by 7 and use the floor function, creating a new variable week. Finally, I can use the

DT[,.(measurement.Sum = sum(measurement)),by=.(userid,week)]

Would this be the fastest way of doing things? (I read about the zoo library but it seems troublesome to switch between data.table and zoo library)

@ulfelder The OP seems to be using data.table, which already includes a `week` function... I'm not clear on why they might need another here...? — Frank, Oct 24 '16 at 15:32
I didn't know that data table contains week! thanks for telling me! — wwl, Oct 24 '16 at 15:33

score 0 · Accepted Answer · edited May 23 '17 at 12:16

0

data.table has a week function:

See, for example, How to get week numbers from dates?

edited May 23 '17 at 12:16

Community

1
1

answered Oct 24 '16 at 15:34

wwl

2,025
2
30
51

Data table aggregate by week and userid

1 Answers1