The mosaic package makes several summary statistic functions (like mean and sd) formula aware.

mean_(x, ..., data = NULL, groups = NULL, na.rm = getOption("na.rm", FALSE))

mean(x, ...)

median(x, ..., data = NULL, groups = NULL, na.rm = getOption("na.rm", FALSE))

range(x, ..., data = NULL, groups = NULL, na.rm = getOption("na.rm", FALSE))

sd(x, ..., data = NULL, groups = NULL, na.rm = getOption("na.rm", FALSE))

max(x, ..., data = NULL, groups = NULL, na.rm = getOption("na.rm", FALSE))

min(x, ..., data = NULL, groups = NULL, na.rm = getOption("na.rm", FALSE))

sum(x, ..., data = NULL, groups = NULL, na.rm = getOption("na.rm", FALSE))

IQR(x, ..., data = NULL, groups = NULL, na.rm = getOption("na.rm", FALSE))

fivenum(x, ..., data = NULL, groups = NULL, na.rm = getOption("na.rm", FALSE))

iqr(x, ..., data = NULL, groups = NULL, na.rm = getOption("na.rm", FALSE))

prod(x, ..., data = NULL, groups = NULL, na.rm = getOption("na.rm", FALSE))

sum(x, ..., data = NULL, groups = NULL, na.rm = getOption("na.rm", FALSE))

favstats(x, ..., data = NULL, groups = NULL, na.rm = TRUE)

quantile(x, ..., data = NULL, groups = NULL, na.rm = getOption("na.rm", FALSE))

var(x, y = NULL, na.rm = getOption("na.rm", FALSE), ..., data = NULL)

cor(x, y = NULL, ..., data = NULL)

cov(x, y = NULL, ..., data = NULL)

## Arguments

x

a numeric vector or a formula

...

data

a data frame in which to evaluate formulas (or bare names). Note that the default is data = parent.frame(). This makes it convenient to use this function interactively by treating the working environment as if it were a data frame. But this may not be appropriate for programming uses. When programming, it is best to use an explicit data argument -- ideally supplying a data frame that contains the variables mentioned.

groups

a grouping variable, typically a name of a variable in data

na.rm

a logical indicating whether NAs should be removed before computing

y

a numeric vector or a formula

## Details

Many of these functions mask core R functions to provide an additional formula interface. Old behavior should be unchanged. But if the first argument is a formula, that formula, together with data are used to generate the numeric vector(s) to be summarized. Formulas of the shape x ~ a or ~ x | a can be used to produce summaries of x for each subset defined by a. Two-way aggregation can be achieved using formulas of the form x ~ a + b or  x ~ a | b. See the examples.

## Note

Earlier versions of these functions supported a "bare name + data frame" interface. This functionality has been removed since it was (a) ambiguous in some cases, (b) unnecessary, and (c) difficult to maintain.

## Examples

mean(HELPrct$age) #> [1] 35.65342 mean( ~ age, data = HELPrct) #> [1] 35.65342 mean( ~ drugrisk, na.rm = TRUE, data = HELPrct) #> [1] 1.887168 mean(age ~ shuffle(sex), data = HELPrct) #> female male #> 35.78505 35.61272 mean(age ~ shuffle(sex), data = HELPrct, .format = "table") #> shuffle(sex) mean #> 1 female 36.26168 #> 2 male 35.46532 # wrap in data.frame() to auto-convert awkward variable names data.frame(mean(age ~ shuffle(sex), data = HELPrct, .format = "table")) #> shuffle.sex. mean #> 1 female 35.74766 #> 2 male 35.62428 mean(age ~ sex + substance, data = HELPrct) #> female.alcohol male.alcohol female.cocaine male.cocaine female.heroin #> 39.16667 37.95035 34.85366 34.36036 34.66667 #> male.heroin #> 33.05319 mean( ~ age | sex + substance, data = HELPrct) #> female.alcohol male.alcohol female.cocaine male.cocaine female.heroin #> 39.16667 37.95035 34.85366 34.36036 34.66667 #> male.heroin #> 33.05319 mean( ~ sqrt(age), data = HELPrct) #> [1] 5.936703 sum( ~ age, data = HELPrct) #> [1] 16151 sd(HELPrct$age)
#> [1] 7.710266
sd( ~ age, data = HELPrct)
#> [1] 7.710266
sd(age ~ sex + substance, data = HELPrct)
#> female.alcohol   male.alcohol female.cocaine   male.cocaine  female.heroin
#>       7.980333       7.575644       6.195002       6.889772       8.035839
#>    male.heroin
#>       7.973568
var(HELPrct\$age)
#> [1] 59.4482
var( ~ age, data = HELPrct)
#> [1] 59.4482
var(age ~ sex + substance, data = HELPrct)
#> female.alcohol   male.alcohol female.cocaine   male.cocaine  female.heroin
#>       63.68571       57.39037       38.37805       47.46896       64.57471
#>    male.heroin
#>       63.57779
IQR(width ~ sex, data = KidsFeet)
#>    B    G
#> 0.75 0.60
iqr(width ~ sex, data = KidsFeet)
#>    B    G
#> 0.75 0.60
favstats(width ~ sex, data = KidsFeet)
#>   sex min    Q1 median    Q3 max     mean        sd  n missing
#> 1   B 8.4 8.875   9.15 9.625 9.8 9.190000 0.4517801 20       0
#> 2   G 7.9 8.550   8.80 9.150 9.5 8.784211 0.4935846 19       0

cor(length ~ width, data = KidsFeet)
#> [1] 0.6410961
cov(length ~ width, data = KidsFeet)
#> [1] 0.4304453
tally(is.na(mcs) ~ is.na(pcs), data = HELPmiss)
#>           is.na(pcs)
#> is.na(mcs) TRUE FALSE
#>      TRUE     2     0
#>      FALSE    0   468
cov(mcs ~ pcs, data = HELPmiss)             # NA because of missing data
#> [1] NA
cov(mcs ~ pcs, data = HELPmiss, use = "complete")  # ignore missing data
#> [1] 13.46433
# alternative approach using filter explicitly
cov(mcs ~ pcs, data = HELPmiss %>% filter(!is.na(mcs) & !is.na(pcs)))
#> [1] 13.46433