How to get values of a numeric column with column name as a parameter in a for loop in R-CodePudding

I have a dataframe with numerous numeric columns that I want to analyze one by one, so I am doing this in a for loop rather than copy/paste the same set of code for each variable. As an example, one of those columns are titled "weight_dry" with values in the 1400 to 1500 (over 180k values in that range). My question is, how would I refer to this parameter so that its value is recognized and used. For example, the 1st two lines of code below does not filter the values of "weight_dry" as the code section below does when use the variable name explicitly. I figure there must be an easy way to do this - I just don't know and can't find a good reference

this doesn't work, even though I have confirmed that the code knows that, for the 1st iteration of j, test_par[j]=weight_dry

all_data2 <- all_data2 %>%
filter(test_par[j] > 0)

whereas, this works just fine:

all_data2 <- all_data2 %>%
filter(weight_dry > 0)

I tried to use eval() and maybe didn't use it right but didn't work for me (so far).

CodePudding user response：

We can just use .data - recommended way to deal with this kind of issue

all_data2 <- all_data2 %>%
            filter(.data[[test_par[j]]] > 0)

or another option is across with all_of

all_data2 <- all_data2 %>%
           filter(across(all_of(test_par[j]), ~ . > 0))

CodePudding user response：

If test_par is a character vector of variable names, we have to convert to symbol before evaluating.

We can use rlang, convert to symbol and evaluate (!!):

library(dplyr)
library(rlang)

all_data2 <- all_data2 %>%
           filter(!!sym(test_par[j]) > 0)