日本語版はこちら (README in Japanese)
Aggregate values by specified groups and adds their group totals
You can install totalize directly from its GitHub repository:
devtools::install_github("ichiromurata/totalize")
Standard R functions provide several methods for data aggregation.
For instance, using datasets::CO2 as an example, you can count data
with table():
table(datasets::CO2$Type, datasets::CO2$Treatment)
#>
#> nonchilled chilled
#> Quebec 21 21
#> Mississippi 21 21Alternatively, you can sum values using tapply(), stats::xtabs(),
tapply(datasets::CO2$uptake, datasets::CO2[c("Type", "Treatment")], FUN=sum)
#> Treatment
#> Type nonchilled chilled
#> Quebec 742 666.8
#> Mississippi 545 332.1
stats::xtabs(uptake~Type+Treatment, data=datasets::CO2)
#> Treatment
#> Type nonchilled chilled
#> Quebec 742.0 666.8
#> Mississippi 545.0 332.1or stats::aggregate().
stats::aggregate(datasets::CO2["uptake"], datasets::CO2[c("Type", "Treatment")], sum)
#> Type Treatment uptake
#> 1 Quebec nonchilled 742.0
#> 2 Mississippi nonchilled 545.0
#> 3 Quebec chilled 666.8
#> 4 Mississippi chilled 332.1However, a common limitation of these functions is that they do not compute subtotals (e.g., by ‘Type’ or ‘Treatment’) and the grand total in a single step. Calculating these different levels of aggregation typically requires additional, separate operations, as the functions are not designed to consolidate them into a unified output.
totalize streamlines this entire process into a single command:
library(totalize)
totalize(datasets::CO2, c(Type, Treatment), val=uptake, asDF=TRUE)
#> Type Treatment uptake
#> 1 all_Type all_Treatment 2285.9
#> 2 all_Type nonchilled 1287.0
#> 3 all_Type chilled 998.9
#> 4 Quebec all_Treatment 1408.8
#> 5 Quebec nonchilled 742.0
#> 6 Quebec chilled 666.8
#> 7 Mississippi all_Treatment 877.1
#> 8 Mississippi nonchilled 545.0
#> 9 Mississippi chilled 332.1Furthermore, totalize can produce a neatly arranged matrix
representation of the results:
totalize(datasets::CO2, row=Type, col=Treatment, val=uptake)
#> Treatment
#> Type all_Treatment nonchilled chilled
#> all_Type 2285.9 1287 998.9
#> Quebec 1408.8 742 666.8
#> Mississippi 877.1 545 332.1When grouping by three or more variables, totalize organizes the
output into a 2-dimensional matrix by creating combined group labels for
clarity:
totalize(datasets::CO2, row=Type, col=c(Treatment, conc), val=uptake)
#> Treatment,conc
#> Type all_Treatment|all_conc all_Treatment|95 all_Treatment|175 all_Treatment|250 all_Treatment|350 all_Treatment|500 all_Treatment|675 all_Treatment|1000 nonchilled|all_conc nonchilled|95 nonchilled|175 nonchilled|250 nonchilled|350 nonchilled|500 nonchilled|675 nonchilled|1000 chilled|all_conc chilled|95 chilled|175 chilled|250 chilled|350 chilled|500 chilled|675 chilled|1000
#> all_Type 2285.9 147.1 267.4 346.5 368.0 370.5 383.4 403 1287 79.7 150.7 194.8 210.8 210.6 216.1 224.3 998.9 67.4 116.7 151.7 157.2 159.9 167.3 178.7
#> Quebec 1408.8 84.4 162.5 215.6 228.5 228.8 237.0 252 742 45.8 90.1 112.2 121.1 118.8 124.5 129.5 666.8 38.6 72.4 103.4 107.4 110.0 112.5 122.5
#> Mississippi 877.1 62.7 104.9 130.9 139.5 141.7 146.4 151 545 33.9 60.6 82.6 89.7 91.8 91.6 94.8 332.1 28.8 44.3 48.3 49.8 49.9 54.8 56.2
totalize(datasets::CO2, row=c(Type, conc), col=Treatment, val=uptake)
#> Treatment
#> Type,conc all_Treatment nonchilled chilled
#> all_Type|all_conc 2285.9 1287.0 998.9
#> all_Type|95 147.1 79.7 67.4
#> all_Type|175 267.4 150.7 116.7
#> all_Type|250 346.5 194.8 151.7
#> all_Type|350 368.0 210.8 157.2
#> all_Type|500 370.5 210.6 159.9
#> all_Type|675 383.4 216.1 167.3
#> all_Type|1000 403.0 224.3 178.7
#> Quebec|all_conc 1408.8 742.0 666.8
#> Quebec|95 84.4 45.8 38.6
#> Quebec|175 162.5 90.1 72.4
#> Quebec|250 215.6 112.2 103.4
#> Quebec|350 228.5 121.1 107.4
#> Quebec|500 228.8 118.8 110.0
#> Quebec|675 237.0 124.5 112.5
#> Quebec|1000 252.0 129.5 122.5
#> Mississippi|all_conc 877.1 545.0 332.1
#> Mississippi|95 62.7 33.9 28.8
#> Mississippi|175 104.9 60.6 44.3
#> Mississippi|250 130.9 82.6 48.3
#> Mississippi|350 139.5 89.7 49.8
#> Mississippi|500 141.7 91.8 49.9
#> Mississippi|675 146.4 91.6 54.8
#> Mississippi|1000 151.0 94.8 56.2To ignore NA values when aggregating, simply pass arguments like
na.rm=TRUE to the aggregation function (FUN):
# Missing values turn the totals into NA
totalize(datasets::penguins, species, island, val=body_mass, FUN=sum)
#> island
#> species all_island Biscoe Dream Torgersen
#> all_species NA NA 460400 NA
#> Adelie NA 163225 206550 NA
#> Chinstrap 253850 NA 253850 NA
#> Gentoo NA NA NA NA
# Ignoring NAs
totalize(datasets::penguins, species, island, val=body_mass, FUN=sum, na.rm=TRUE)
#> island
#> species all_island Biscoe Dream Torgersen
#> all_species 1437000 787575 460400 189025
#> Adelie 558800 163225 206550 189025
#> Chinstrap 253850 NA 253850 NA
#> Gentoo 624350 624350 NA NAIn many datasets, some records may not be assigned to a category in a grouping variable (i.e., they have NA values). While these “unclassified” items don’t belong to a specific group, they are essential for calculating accurate overall totals.
A common pitfall is to calculate a grand total by summing previously
calculated group totals. This approach incorrectly excludes the
unclassified items. To ensure accuracy, totalize always re-aggregates
from the original, unfiltered data when computing subtotals and the
grand total, thereby correctly including all entries.
For instance, in the
palmerpenguins
dataset, the sum of “female” and “male” does not equal the total number
of penguins because some have an unspecified sex. totalize accounts
for this discrepancy:
# Correctly count the total that does not match the sum of female + male
totalize(datasets::penguins, species, sex)
#> sex
#> species all_sex female male
#> all_species 344 165 168
#> Adelie 152 73 73
#> Chinstrap 68 34 34
#> Gentoo 124 58 61
# `addmargins()` excludes the unclassified data
xtabs(~species+sex, data=datasets::penguins) |>
addmargins()
#> sex
#> species female male Sum
#> Adelie 73 73 146
#> Chinstrap 34 34 68
#> Gentoo 58 61 119
#> Sum 165 168 333totalize can also produce a single-column matrix when only row
groupings are specified:
totalize(datasets::CO2, row=c(Type, Treatment), val=uptake)
#>
#> Type,Treatment uptake
#> all_Type|all_Treatment 2285.9
#> all_Type|nonchilled 1287.0
#> all_Type|chilled 998.9
#> Quebec|all_Treatment 1408.8
#> Quebec|nonchilled 742.0
#> Quebec|chilled 666.8
#> Mississippi|all_Treatment 877.1
#> Mississippi|nonchilled 545.0
#> Mississippi|chilled 332.1To obtain a single-row matrix, you can easily transpose the result using
the t() function:
totalize(datasets::CO2, row=c(Type, Treatment), val=uptake) |> t()
#> Type,Treatment
#> all_Type|all_Treatment all_Type|nonchilled all_Type|chilled Quebec|all_Treatment Quebec|nonchilled Quebec|chilled Mississippi|all_Treatment Mississippi|nonchilled Mississippi|chilled
#> uptake 2285.9 1287 998.9 1408.8 742 666.8 877.1 545 332.1The row, col, and val parameters support ‘non-standard evaluation’
(NSE). This means you can refer to column names directly, without
needing to enclose them in quotes.
This package provides an awayfromzero() method for rounding, which
rounds halves away from
zero.
awayfromzero() is designed to address floating-point precision issues
by dynamically scaling its rounding tolerance based on the magnitude of
the input number.
This rounding function can be conveniently chained with other operations
using the pipe operator |>.
totalize(datasets::CO2, row=c(Treatment, conc), col=Type, val=uptake, FUN=mean) |>
awayfromzero(digits = 1)
#> Type
#> Treatment,conc all_Type Quebec Mississippi
#> all_Treatment|all_conc 27.2 33.5 20.9
#> all_Treatment|95 12.3 14.1 10.5
#> all_Treatment|175 22.3 27.1 17.5
#> all_Treatment|250 28.9 35.9 21.8
#> all_Treatment|350 30.7 38.1 23.3
#> all_Treatment|500 30.9 38.1 23.6
#> all_Treatment|675 32.0 39.5 24.4
#> all_Treatment|1000 33.6 42.0 25.2
#> nonchilled|all_conc 30.6 35.3 26.0
#> nonchilled|95 13.3 15.3 11.3
#> nonchilled|175 25.1 30.0 20.2
#> nonchilled|250 32.5 37.4 27.5
#> nonchilled|350 35.1 40.4 29.9
#> nonchilled|500 35.1 39.6 30.6
#> nonchilled|675 36.0 41.5 30.5
#> nonchilled|1000 37.4 43.2 31.6
#> chilled|all_conc 23.8 31.8 15.8
#> chilled|95 11.2 12.9 9.6
#> chilled|175 19.5 24.1 14.8
#> chilled|250 25.3 34.5 16.1
#> chilled|350 26.2 35.8 16.6
#> chilled|500 26.7 36.7 16.6
#> chilled|675 27.9 37.5 18.3
#> chilled|1000 29.8 40.8 18.7