Tally across columns with variable condition in r-CodePudding

I am trying to tally across columns of a dataframe with values that exceed a corresponding limit variable. Here is a similar problem but for each point, the condition may change so rowSums is not an option. I am interested in either modifications to the solution I have attempted below, any more efficient methods, and any simpler methods which help to solve this issue.

library(tidyverse)

set.seed(234)

ex_dat <- tibble(a = rnorm(n = 6, mean = 1),
                 a_lim = 0.75,
                 b = rnorm(n = 6, mean = 0.5),
                 b_lim = 0.333,
                 c = rnorm(n = 6, mean = 1.5),
                 c_lim = 1.0,
                 d = rnorm(n = 6, mean = 1.5),
                 d_lim = 1.25)

ex_dat %>% 
rowwise() %>% 
mutate(tally = sum(map_lgl(.x = c("a","b","c","d"),
.f = ~(noquote(.x) > noquote(paste0(.x,"_lim")))), na.rm = T))

For instance, the desired outcome would be that the 'tally' column here reads 4 in the first row, given that all 4 conditions are met. Then, it should read 2 for the second row, given that only 2 limits were exceeded.

I have tried a few variations on this approach here, using the bang-bang operator to force-evaluate the variables in the .f argument to the map function. So far, this attempt and silent-failure seems to be the closest and most sensible as far as I can wrap my head around. Evidently, I don't have a very firm grasp on non-standard evaluation, so my attempts with !! and sym() didn't take me far.

Again, if anyone sees that I am going about this problem in a roundabout or inefficient way, I would welcome redirection. Thank you.

CodePudding user response：

Here is a slightly different way to approach your problem. It relies on the "trick" that inequalities evalute to 1 if true, and 0 if false. So you can evaluate a, b, c, and d against their limits, and sum the evaluation of the 4 inequalities.

library(dplyr)

 ex_dat_tally<-ex_dat %>%
     mutate(tally=(a>a_lim) (b>b_lim) (c>c_lim) (d>d_lim))

CodePudding user response：

I find this base R approach intuitive, easy and it should also be fast since we are working on dataframe directly using vectorised approach.

lim_cols <- grep('lim', names(ex_dat), value = TRUE)
cols <- sub('_lim', '', lim_cols)
ex_dat$tally <- rowSums(ex_dat[cols] > ex_dat[lim_cols])
ex_dat

#      a a_lim       b b_lim     c c_lim     d d_lim tally
#   <dbl> <dbl>   <dbl> <dbl> <dbl> <dbl> <dbl> <dbl> <dbl>
#1  1.66   0.75  0.709  0.333 1.47      1 2.03   1.25     4
#2 -1.05   0.75 -2.54   0.333 2.01      1 1.52   1.25     2
#3 -0.499  0.75  0.0131 0.333 2.49      1 1.71   1.25     2
#4  2.47   0.75 -0.588  0.333 1.80      1 2.52   1.25     3
#5  2.46   0.75  0.558  0.333 0.570     1 1.91   1.25     3
#6  1.14   0.75  1.60   0.333 1.58      1 0.795  1.25     3

Here we create two group of columns lim_cols and cols.

lim_cols
#[1] "a_lim" "b_lim" "c_lim" "d_lim"

cols
#[1] "a" "b" "c" "d"

Compare them pairwise and use rowSums to calculate how many of them satisfy the condition.