Using summarise in matsbyname

library(dplyr)
#> 
#> Attaching package: 'dplyr'
#> The following objects are masked from 'package:stats':
#> 
#>     filter, lag
#> The following objects are masked from 'package:base':
#> 
#>     intersect, setdiff, setequal, union
library(matsbyname)
library(tibble)

Introduction

matsbyname functions in which operands are specified in a ... argument are ambiguous when applied to a data frame. But there is an argument (.summarise) that signals intention, allowing the ambiguous functions to be used flexibly with data frames.

“Normal” functions

For normal functions, such as + and mean(), there is no ambiguity about their operation in a data frame.

df <- tibble::tribble(~x, ~y, ~z, 
                       1,  2,  3, 
                       4,  5,  6)
# Typically, operations are done across rows.
df %>% 
  dplyr::mutate(
    a = x + y + z,
    b = rowMeans(.)
  )
#> # A tibble: 2 × 5
#>       x     y     z     a     b
#>   <dbl> <dbl> <dbl> <dbl> <dbl>
#> 1     1     2     3     6     2
#> 2     4     5     6    15     5

To perform the same operations down columns, use dplyr::summarise().

df %>% 
  dplyr::summarise(
    x = sum(x), 
    y = sum(y), 
    z = sum(z)
  )
#> # A tibble: 1 × 3
#>       x     y     z
#>   <dbl> <dbl> <dbl>
#> 1     5     7     9
df %>% 
  dplyr::summarise(
    x = mean(x), 
    y = mean(y), 
    z = mean(z)
  )
#> # A tibble: 1 × 3
#>       x     y     z
#>   <dbl> <dbl> <dbl>
#> 1   2.5   3.5   4.5

matsbyname::sum_byname()

What does matsbyname::sum_byname() mean for a data frame? Will it give sums across rows (as +), or will it give sums down columns (as summarise())? This ambiguity is present for all *_byname() functions in which operands are specified via the ... argument, including matrixproduct_byname(), hadamardproduct_byname(), mean_byname(), etc.

To resolve the ambiguity, use the .summarise argument. The default value of .summarise is FALSE, meaning that the functions normally operate across rows. If you want to perform the action down columns, set .summarise = TRUE.

df %>% 
  dplyr::mutate(
    a = sum_byname(x, y, z), 
    b = mean_byname(x, y, z)
  )
#> # A tibble: 2 × 5
#>       x     y     z     a     b
#>   <dbl> <dbl> <dbl> <dbl> <dbl>
#> 1     1     2     3     6     2
#> 2     4     5     6    15     5
df %>% 
  dplyr::summarise(
    x = sum_byname(x, .summarise = TRUE) %>% unlist(), 
    y = sum_byname(y, .summarise = TRUE) %>% unlist(), 
    z = sum_byname(z, .summarise = TRUE) %>% unlist()
  )
#> # A tibble: 1 × 3
#>       x     y     z
#>   <dbl> <dbl> <dbl>
#> 1     5     7     9

Summary

The .summarise argument broadens the range of applicability for many matsbyname functions, especially when used with data frames. The default is .summarise = FALSE, meaning that operations will be performed across columns. Set .summarise = TRUE argument to signal intent to perform operations down a column.