mutate at - count the number of times specified columns exceed some specified number, - apply

Question

This is a real-world problem but I am describing the reproducible iris example:
I want to count, for each row, the number of times columns with column names containing "Sepal", exceed a number 5. I want to assign the new result in a new column. I want to use dplyr for this task. My attempt is the following:
iris %>% mutate_at(vars(contains("Sepal")),list(greater_than_5=~apply(.,1,function(x) sum(x>5))))
However, I get an error:
dim(X) must have a positive length```
Any ideas?

Answer 1

Without seeing an example of what you want your output to look like and/or explanation, it's hard to determine exactly what you're looking for, so here are two possible solutions (including the one I mentioned in the comments above). The first I think will be unsatisfactory since your count will be the same for all rows because the tibble is static. The second uses R's bread and butter--a tidy (long form) data structure--to count only those "columns" that have non-NA values. There we add a row id that we can group by later, pivot to a tidy form, filter out NAs, count the number of column/parameter names that contain your word of interest, and the pivot back to your original wide form tibble. We need to wrap the value column in list form since the datatypes differ across the different parameters--you could convert everything to a character, of course, but you'll see this lets us recover the original types after unnesting.
library(tidyverse)
min_to_count = 5
title_word = "Sepal"
iris %>% mutate(name_count = ifelse(sum(str_detect(names(.),title_word)) > min_to_count, sum(str_detect(names(.),title_word)), NA_real_))
iris %>%
mutate(id = row_number()) %>%
pivot_longer(
-id,
names_to = "parameter", values_to = "value",
values_ptypes = list(value = list())
) %>%
filter(!is.na(unlist(value))) %>%
group_by(id) %>%
mutate(
name_count = ifelse(
sum(str_detect(parameter, title_word)) > min_to_count,
sum(str_detect(parameter, title_word)), NA_real_
)
) %>%
pivot_wider(
names_from = "parameter", values_from = "value"
) %>%
unnest_legacy()