Joining tables and applying functions to columns with the same name in R and tidyverse - r

Question

I am looking to join tables with customer id (easy enough) but then I want to multiply the columns to get updated values.
Customer_Week_1<-data.frame(First_name=c("John","Mary","David","Paul"),
Last_name=c("Jackson","Smith","Williams", "Zimmerman"),
Factor_1=c(2,5,8,9),
Factor_2=c(.5,.5,.75,.75),
Factor_3=c(0,1,2,3))
Customer_Week_2<-data.frame(First_name=c("John","Mary","David","Paul"),
Last_name=c("Jackson","Smith","Williams", "Zimmerman"),
Factor_1=c(3,7,1,7),
Factor_2=c(.51,.65,.72,.4),
Factor_3=c(1,2,3,4))
Customer_week3<-Customer_Week_1%>%
left_join(Customer_Week_2, by = c("First_name","Last_name"))
The expected results can be found by in a vector by just
Customer_week3_expected<-Customer_Week_1[,3:5]*Customer_Week_2[,3:5]
And I know I can just manually type out every column. But I have dozens of columns and need to make this code as easy to follow as possible.
I also know that I can just bind the results vector to
Customer_week3<-Customer_Week_1%>%
left_join(Customer_Week_2, by = c("First_name","Last_name"))%>%
select(1:2)
But that does not look like best practice to me, and I would rather this be done with a join some way to ensure everything lines up when I am iterating over the customers(tables)

Answer 1

Assuming I understand the output you're trying to get, I can think of two methods. If you know that the names are in the first two columns and are the same in both data frames (this might not be the case in real life), you can use the same multiplication operation you tried above, bound to the first two columns of either of the data frames.
cbind(Customer_Week_1[1:2], Customer_Week_1[-1:-2] * Customer_Week_2[-1:-2])
#> First_name Last_name Factor_1 Factor_2 Factor_3
#> 1 John Jackson 6 0.255 0
#> 2 Mary Smith 35 0.325 2
#> 3 David Williams 8 0.540 6
#> 4 Paul Zimmerman 63 0.300 12
Or you can be more verbose but maybe more flexible, and eshape to a long data frame, then do a grouped operation to summarize products for each person and factor. Starting from the join you have above:
library(dplyr)
library(tidyr)
Customer_week3 <- Customer_Week_1 %>%
left_join(Customer_Week_2, by = c("First_name", "Last_name"))
Make long-shaped data, separate the Factor_1.x into Factor_1 and x, and make products as your summary calculation.
products <- Customer_week3 %>%
gather(key = factor, value = value, -First_name, -Last_name) %>%
separate(factor, into = c("factor", "week"), sep = "\\.") %>%
group_by(First_name, Last_name, factor) %>%
summarise(value = prod(value))
head(products)
#> # A tibble: 6 x 4
#> # Groups: First_name, Last_name [2]
#> First_name Last_name factor value
#> <fct> <fct> <chr> <dbl>
#> 1 David Williams Factor_1 8
#> 2 David Williams Factor_2 0.54
#> 3 David Williams Factor_3 6
#> 4 John Jackson Factor_1 6
#> 5 John Jackson Factor_2 0.255
#> 6 John Jackson Factor_3 0
If you need to get back to a wide format, spread back.
products %>%
spread(key = factor, value = value)
#> # A tibble: 4 x 5
#> # Groups: First_name, Last_name [16]
#> First_name Last_name Factor_1 Factor_2 Factor_3
#> <fct> <fct> <dbl> <dbl> <dbl>
#> 1 David Williams 8 0.54 6
#> 2 John Jackson 6 0.255 0
#> 3 Mary Smith 35 0.325 2
#> 4 Paul Zimmerman 63 0.3 12

Answer 2

Similar to #camille's reshaping, but in data.table (and disregarding Customer_week3):
library(data.table)
# long format
long = rbindlist(list(Customer_Week_1, Customer_Week_2), id=TRUE)
# aggregate
long[, lapply(.SD, prod), by=.(First_name, Last_name), .SDcols=patterns("^Factor")]
First_name Last_name Factor_1 Factor_2 Factor_3
1: John Jackson 6 0.255 0
2: Mary Smith 35 0.325 2
3: David Williams 8 0.540 6
4: Paul Zimmerman 63 0.300 12
Going longer (again as seen in #camille's answer) might also make sense, so as to avoid repeatedly fiddling with names of Factor_* columns:
longer = melt(long, meas=patterns("^Factor")) # analogous to gather
longer[, .(value = prod(value)), by=.(First_name, Last_name, variable)]