Add formatting functionality to package

[Yeon-Choi-git]

Hi @CarlosPoses,

I made some minor updates to your mask_vector() so that:

1. it can take an input that is a dataset in a wide format (like one row data set instead of a vector)
2. it always return a character string consistently, even when there is no number that were masked.
   • add an option to add a big number separator
   • show trailing zeros until the intended decimals

Please see an example.

testdat <- data.table::data.table(var1 = 4, var2= 2, var3 = 4000)

testdat #data.table object
  var1  var2  var3
 <num> <num> <num>
     4     2  4000

mask_vector(testdat, 
            threshold = 5, 
            rounding_digits = 2,
            big_mark_seperator = ",",
            output_warnings = TRUE, 
            percentage = FALSE,
            total = NULL)
         var1            var2            var3 
      "[1-4]"         "[1-4]" "[3,998-4,004]" 

mask_vector(testdat, 
            threshold = 5, 
            rounding_digits = 2,
            big_mark_seperator = ",",
            output_warnings = TRUE, 
            percentage = TRUE,
            total = NULL)
         var1            var2            var3 
"[0.02-0.10]"   "[0.02-0.10]" "[99.80-99.95]" 

[CarlosPoses]

Hi Yeon,

Thanks a lot! Nice features.

• For 2 (formatting), I will merge without changes.

• For 1 (accepting dataframe as input), I would suggest adding a small wrapper to support this instead of allowing a dataframe or datatable object as input for the mask_vector() function.

  My rationale: as the package is now, there're two masking functions.

  1. Mask vector(): takes vector as input and masks that vector. I think it's good to keep this for vector input only.
  2. Mask_vector_wrapper(). Takes DescriptiveTable output (which is just a particular dataframe structure created with this package), identifies which vectors need to be masked from this table and applies the masking.

  More generally depending on the data.frame or data.table structure you want to use the function with, which rows and/or columns combinations correspond to the vector that need to be masked will be different. It totally makes sense to add your specific case, but I would add one different wrapper for each different data.frame or data.table. For instance I would call this wrapper mask_vector_row_tables(), having a third function.

  Would you agree? It's a small thing but I think it would help in maintaining or scalating. For instance if someone else wants to use this function for a different data.frame structure, it's clear they need to write a new wrapper that specifies what are vectors to be masked. I am happy to write the code. Might as well through an error for non-vector inputs and add some documentation about this wrappers idea.

Carlos

[Yeon-Choi-git]

Yup make sense wil do that. thank you!

[Yeon-Choi-git]

Added a condition to also return formatting values when a vector contain 0 value to mask.

[Yeon-Choi-git]

hi @CarlosPoses,

I made a new push including:

1. added a condition in masking_vector() to return formatted numbers also when the input vector contains no values to mask.

> # Nothing to mask
> mask_vector(c(0,5,1000),
+             threshold = 5,
+             rounding_digits = 2,
+             big_mark_seperator = ",",
+             output_warnings = TRUE,
+             percentage = FALSE,
+             total = NULL
+ )
[1] "0"     "5"     "1,000"

2. a wrapper function as you suggested - which can also perform the secondary masking per row.

> testdat <- data.table::data.table(var1 = c(1, 4), 
+                                   var2 = c(0, 2), 
+                                   var3 = c(10, 3), 
+                                   var4 = c(3000, 4000)
+ )
> testdat
    var1  var2  var3  var4
   <num> <num> <num> <num>
1:     1     0    10  3000
2:     4     2     3  4000
> mask_row_table(input_row_table = testdat ,
+                 maskcols = c("var2", "var3", "var4"),
+                 threshold = 5,
+                 rounding_digits = 2,
+                 big_mark_seperator = ",",
+                 output_warnings = FALSE,
+                 percentage = FALSE,
+                 total = NULL)
    var1   var2   var3          var4
   <num> <char> <char>        <char>
1:     1      0     10         3,000
2:     4  [1-4]  [1-4] [3,997-4,003]

[LindaNab]

@CarlosPoses can you look at the above please? Preferably we would merge this into main

[CarlosPoses]

This crashes for some values of is_expection = TRUE. Concretely formatC seems to not support NA and NaNs

[CarlosPoses]

Hi @Yeon-Choi-git ,

I just look at it, sorry for the long waiting time. The function crashes when running the tests (you can source it tests/testthat/test-interval-masking.R' and it's related to the new formatting.

1. Is the expected output of mask_vector always a character now? This seems fine to me. I assumed so, so that's what a test that was failing reflects now.
2. The error seems that the new formatting functionality doesn't handle well the exception values. Concretely the problem is caused by NA and NaN values now being an acceptable input to formatC. Could you take a look?

Thanks!

Carlos