left_roll: Left or right rolling join

In decisionpatterns/data.table.plus: Extensions and Utilities for 'data.table'

Description

Perform rolling join

Usage

left_roll(x, y, on, roll, ...)

Arguments

x,	y data.table
on	character vector of join columns. The last is the join column
roll	direction and allowed size of the roll. see ?data.table.
...	additional arguments passed to setprefix (unimplemented)

Details

It is best to think of x as the m

Duplicated y's

1. It is often expected that nrow(results) equals nrow(x). This is not true when are duplicated rows in y; it maybe important to the application to dedupe the rows of y before joining?

2. When deduping y, ensure that you are taking the correct duplicate. 'unique“ retains the first occerence.

Value

a data.table object with

Note

• The name roll comes from the data.table/ts rolling joins

• zip is also an interesting name – zipping on a fields with roll telling what to do first

Author(s)

Decision Patterns

See Also

• setprefix

• dplyr::left_join()

• data.table::[.data.table

Examples

set.seed(1)
x <- data.table(
         letter=sample(letters[1:2],10, replace=TRUE)
       , year_x=sample( 2000:2010, 10, replace=TRUE )
     )
x <- x[ order(letter, year_x) ]
#' x <- unique(x)

y <- data.table(
         letter=sample(letters[1:2],5, replace=TRUE)
       , year_y=sample( 2000:2010, 5, replace=TRUE )
     )
y <- y[ order(letter, year_y) ]
y <- unique(y)
y

# on=c('letter','year')
on=c('letter',year_x='year_y')
roll=Inf


ret <- left_roll(x,y,on=on,roll=roll)
ret[ letter=="a"]; x[letter=="a"]; y[ letter=="a"]
ret[ letter=="b"]; x[letter=="b"]; y[ letter=="b"]

decisionpatterns/data.table.plus documentation built on June 15, 2020, 10:26 p.m.