s*********e
发帖数: 1051 | |
D**u
发帖数: 288 | 2 Very impressive!
How long does this step take?
dt <- data.table(df) |
D**u
发帖数: 288 | 3 I think Sqldf (working with SQLite) still have the edge in big data over
data.table. |
s*********e
发帖数: 1051 | 4 Sqldf is slow, period
【在 D**u 的大作中提到】
: I think Sqldf (working with SQLite) still have the edge in big data over
: data.table.
|
D**u
发帖数: 288 | 5 Ok, I am going to try hdf5 + data.table combination, compare to rsqlite+
sqldf. That will be the optimal way I can think of now. |
s*********e
发帖数: 1051 | 6 data.table is 100X faster than sqldf() for row extraction from a data.frame.
http://statcompute.wordpress.com/2013/01/01/efficiecy-of-extrac |
w******4
发帖数: 488 | |
