The R4DS book has the following code block:
library(tidyverse)
by_age2 <- gss_cat %>%
filter(!is.na(age)) %>%
count(age, marital) %>%
group_by(age) %>%
mutate(prop = n / sum(n))
Is there a simple equivalent to this code in base R? The filter can be replaced with gss_cat[!is.na(gss_cat$age),], but after that I run in to trouble. It's clearly a job for by, tapply, or aggregate, but I've not been able to find the right way. by(gss_2, with(gss_2, list(age, marital)), length) is a step in the right direction, but the output is awful.
We could use proportions on the table output after subsetting to remove the NA (complete.cases) and selecting the columns
The data is from forcats package. So, load the package and get the data
library(forcats)
data(gss_cat)
Use the table/proportions as mentioned above
by_age2_base <- proportions(table(subset(gss_cat, complete.cases(age),
select = c(age, marital))), 1)
-output
head(by_age2_base, 3)
marital
age No answer Never married Separated Divorced Widowed Married
18 0.000000000 0.978021978 0.000000000 0.000000000 0.000000000 0.021978022
19 0.000000000 0.939759036 0.000000000 0.012048193 0.004016064 0.044176707
20 0.000000000 0.904382470 0.003984064 0.007968127 0.000000000 0.083665339
-compare with the OP's output
head(by_age2, 3)
# A tibble: 3 x 4
# Groups: age [2]
age marital n prop
<int> <fct> <int> <dbl>
1 18 Never married 89 0.978
2 18 Married 2 0.0220
3 19 Never married 234 0.940
If we need the output in 'long' format, convert the table to data.frame with as.data.frame
by_age2_base_long <- subset(as.data.frame(by_age2_base), Freq > 0)
Or another option is aggregate/ave (use R 4.1.0)
subset(gss_cat, complete.cases(age), select = c(age, marital)) |>
{\(dat) aggregate(cbind(n = age) ~ age + marital,
data = dat, FUN = length)}() |>
transform(prop = ave(n, age, FUN = \(x) x/sum(x)))