Merge and Aggregate by multiple colums in r? - r

Question

I have 2 tables. Below are the sample tables and the desired output.
Table1:
Start Date End Date Country
2017-01-04 2017-01-06 id
2017-02-13 2017-02-15 ng
Table2:
Transaction Date Country Cost Product
2017-01-04 id 111 21
2017-01-05 id 200 34
2017-02-14 ng 213 45
2017-02-15 ng 314 32
2017-02-18 ng 515 26
Output:
Start Date End Date Country Cost Product
2017-01-04 2017-01-06 id 311 55
2017-02-13 2017-02-15 ng 527 77
The problem is to merge two tables when transaction date lies in between start date and end date & country matches. And add the values of cost and product.

Answer 1

This calls for fuzzyjoins. Below are 2 examples.
Using dplyr and fuzzyjoin packages:
fuzzy_left_join(df1, df2,
c("Country" = "Country",
"Start_Date" = "Transaction_Date",
"End_Date" = "Transaction_Date"),
list(`==`, `<=`,`>=`)) %>%
group_by(Country.x, Start_Date, End_Date) %>%
summarise(Cost = sum(Cost),
Product = sum(Product))
# A tibble: 2 x 5
# Groups: Country.x, Start_Date [?]
Country.x Start_Date End_Date Cost Product
<chr> <date> <date> <int> <int>
1 id 2017-01-04 2017-01-06 311 55
2 ng 2017-02-13 2017-02-15 527 77
Using data.table:
library(data.table)
dt1 <- data.table(df1)
dt2 <- data.table(df2)
dt2[dt1, on=.(Country = Country,
Transaction_Date >= Start_Date,
Transaction_Date <= End_Date),
.(Cost = sum(Cost), Product = sum(Product)),
by=.EACHI]
data:
df1 <- structure(list(Start_Date = structure(c(17170, 17210), class = "Date"),
End_Date = structure(c(17172, 17212), class = "Date"), Country = c("id",
"ng")), row.names = c(NA, -2L), class = "data.frame")
df2 <- structure(list(Transaction_Date = structure(c(17170, 17171, 17211,
17212, 17215), class = "Date"), Country = c("id", "id", "ng",
"ng", "ng"), Cost = c(111L, 200L, 213L, 314L, 515L), Product = c(21L,
34L, 45L, 32L, 26L)), row.names = c(NA, -5L), class = "data.frame")

Answer 2

Not sure if you can use any of the merge operation here but one way using mapply is to subset the rows based on the condition and take the sum of Product and Cost columns.
df1[c("Cost", "Product")] <- t(mapply(function(x, y, z) {
inds <- df2$Transaction_Date >= x & df2$Transaction_Date <= y & df2$Country == z
c(sum(df2$Cost[inds]), sum(df2$Product[inds]))
},df1$Start_Date, df1$End_Date, df1$Country))
df1
# Start_Date End_Date Country Cost Product
#1 2017-01-04 2017-01-06 id 311 55
#2 2017-02-13 2017-02-15 ng 527 77