Iterate by month with lubridate and merge combined - r

Question

I am trying to write a function that merges based on two columns both found in two dataframes. One of the columns is an identifier string and the other is a date.
The first df ("model") includes identifiers, starting dates, and some other relevant info.
The second df ("futurevalues") is a melted df that includes the identifier, multiple months for each identifier, and the relevant value for each identifier-month pair.
I would like to merge values for each identifier based on a certain period of time in the future. So for instance, for Identifier= Mary and starting month="2005-01-31" in "model" I would like to pull in the relevant value for the next month and 11 more months after (so 12 data points for Mary for months starting month+1:starting month+12).
I can merge my dfs by the two columns to get the as-of date value (see below), but this isn't what I need.
testmerge=merge(model,futurevalues,by=c("month","identifier"),all=TRUE)
To solve this, I am trying to use the lubridate date functions. For instance, the function below will allow me to enter a month (and then lapply across the df maybe) to get the values for each of the starting months (which vary across the df, meaning it's not a standard time period across the entire thing).
monthiterate=function (x) {
x %m+% months(1:12)
}
Thanks a lot for your help.
EDIT: adding toy data (first one is model, second one is futurevalues)
structure(list(month = structure(c(12814, 12814, 12814, 12814,
12814, 12814, 12814, 12814, 12814, 12814), class = "Date"), identifier = structure(c(1L,
3L, 2L, 4L, 5L, 7L, 8L, 6L, 9L, 10L), .Label = c("AB1", "AC5",
"BB9", "C99", "D81", "GG8", "Q11", "R45", "ZA1", "ZZ9"), class = "factor"),
value = c(0.831876072999969, 0.218494398256579, 0.550872926656984,
1.81882711231324, -0.245597705276932, -0.964277509916354,
-1.84714556574606, -0.916239506529079, -0.475649743547525,
-0.227721186387637)), .Names = c("month", "identifier", "value"
), class = "data.frame", row.names = c(NA, 10L))
structure(list(identifier = structure(c(1L, 3L, 2L, 4L, 5L, 7L,
8L, 6L, 9L, 10L), .Label = c("AB1", "AC5", "BB9", "C99", "D81",
"GG8", "Q11", "R45", "ZA1", "ZZ9"), class = "factor"), month = structure(c(12814,
13238, 12814, 12814, 12964, 12903, 12903, 12842, 13148, 13148
), class = "Date"), futurereturns = c(-0.503033205660682, 1.22446988772542,
-0.825490985851348, 1.03902417581908, 0.172595565260429, 0.894967582911769,
-0.242324006922964, 0.415520398113024, -0.734437328639625, 2.64184935856802
)), .Names = c("identifier", "month", "futurereturns"), class = "data.frame", row.names
= c(NA, 10L))

Answer 1

You need to create a table of all the combinations of ID and month that you want. Starting with a table of each ID and their starting month:
library(lubridate)
set.seed(1834)
# 3 people, each with a different starting month
x <- data.frame(id = sample(LETTERS, 3)
, month = ymd("2005-01-01") + months(sample(0:11, 3)) - days(1))
> x
id month
1 D 2005-03-31
2 R 2005-07-31
3 Y 2005-02-28
Now add rows for the following two months, per ID. I use dplyr for this kind of thing.
library(dplyr)
y <- x %>%
rowwise %>%
do(data.frame(id = .$id
, month = seq(.$month + days(1)
, by = "1 month"
, length.out = 3) - days(1)))
> y
Source: local data frame [9 x 2]
Groups: <by row>
id month
1 D 2005-03-31
2 D 2005-04-30
3 D 2005-05-31
4 R 2005-07-31
5 R 2005-08-31
6 R 2005-09-30
7 Y 2005-02-28
8 Y 2005-03-31
9 Y 2005-04-30
Now you can use merge() (or left_join() from dplyr) to retrieve the rows you want from the full dataset.