I'm working on converting Stata code to R. There's a snippet of code that creates a new variable and adds the column value if it meets specific parameters. For example, if a cell is greater than 0 and less than or equal to 3, that value would be added to newvar
gen newvar=0
local list a b c
foreach x of local list{
qui replace newvar=newvar+`x' if `x'>0 & `x'<=3
}
set.seed(5)
dat <- data.frame(a = rnorm(5), b = rnorm(5), c = rnorm(5))
Desired Output
A tidyverse approach
library(dplyr)
set.seed(5)
dat <- data.frame(a = rnorm(5), b = rnorm(5), c = rnorm(5))
conditional_sum <- function(x,a = 0,b = 3){
sum(x[x > a & x <= b],na.rm = TRUE)
}
dat %>%
rowwise() %>%
mutate(newvar = conditional_sum(c_across()))
# A tibble: 5 x 4
# Rowwise:
a b c newvar
<dbl> <dbl> <dbl> <dbl>
1 -0.841 -0.603 1.23 1.23
2 1.38 -0.472 -0.802 1.38
3 -1.26 -0.635 -1.08 0
4 0.0701 -0.286 -0.158 0.0701
5 1.71 0.138 -1.07 1.85
Replace the elements that are not satisfying the condition to NA and get the rowSums on the rest of the elements to create the 'newvar'
dat$newvar <- rowSums(NA^(dat <=0|dat >=3)*dat, na.rm = TRUE)
-output
> dat
a b c newvar
1 -0.84085548 -0.6029080 1.2276303 1.22763034
2 1.38435934 -0.4721664 -0.8017795 1.38435934
3 -1.25549186 -0.6353713 -1.0803926 0.00000000
4 0.07014277 -0.2857736 -0.1575344 0.07014277
5 1.71144087 0.1381082 -1.0717600 1.84954910
a & b without dropping c, how would the code be updated?
dat[c("a", "b")] instead of dat in the code i.e. rowSums(NA^(dat[c("a", "b")] <=0|dat[c("a", "b")]>=3)*dat[c("a", "b")], na.rm = TRUE)A common way to perform rowwise operations is using the apply function. E.g.:
dat$newvar <- apply(dat, 1, \(r) sum(r[r > 0 & r <= 3]))
Read as: Apply a function to every row of dat. The function takes a vector r, and sums the elements of r which satisfy the criterio.
Results in
a b c newvar
1 -0.84085548 -0.6029080 1.2276303 1.22763034
2 1.38435934 -0.4721664 -0.8017795 1.38435934
3 -1.25549186 -0.6353713 -1.0803926 0.00000000
4 0.07014277 -0.2857736 -0.1575344 0.07014277
5 1.71144087 0.1381082 -1.0717600 1.84954910
