Make a function applicable to grouped data.
Value
A function like f with a new argument group. This accepts a factor to
split all other arguments in f into groups before applying f to each
group and combining the results. It is similar to ave(), but more general.
See also
Other operators:
balanced(),
quantity_index()
Examples
p1 <- price6[[3]]
p0 <- price6[[2]]
q1 <- quantity6[[3]]
q0 <- quantity6[[2]]
# Calculate Tornqvist weights for two groups
f <- factor(rep(letters[1:2], each = 3))
tornqvist_weights <- grouped(index_weights("Tornqvist"))
tornqvist_weights(p1, p0, q1, q0, group = f)
#> [1] 0.18247009 0.33481172 0.48271819 0.12865397 0.81833234 0.05301369
# Calculate a mean like ave(), but with weights
x <- 1:6
w <- c(1:5, NA)
grouped_mean <- grouped(\(x, w) geometric_mean(x, w, na.rm = TRUE))
grouped_mean(x, w, group = f)
#> [1] 2.182247 2.182247 2.182247 4.527921 4.527921 4.527921
# Redistribute weights
w1 <- c(2, 4)
w2 <- 1:6
harmonic_mean(mapply(harmonic_mean, split(x, f), split(w2, f)), w1)
#> [1] 3.333333
wr <- grouped(scale_weights)(w2, group = f) * w1[f]
harmonic_mean(x, wr)
#> [1] 3.333333