← SASnRConvert Between Character and Numeric Variables

Introduction

Type conversion between character and numeric is a constant need in clinical data — strings come in from CDISC specs, numbers come back out for analysis.
Two directions to know: character to numeric for analysis, numeric to character for report layout.
In both directions, invalid values (e.g. "ABC" in a numeric column) need to become missing rather than crashing the program.
SAS uses input(char, informat.) for character-to-numeric and put(num, format.) for numeric-to-character.
R uses as.numeric() and as.character() directly in both tidyverse and base dialects.
This lesson demonstrates both conversions on a mixed dataset that includes one invalid string ("ABC") so we can see how each language handles the edge case.

Convert Between Character and Numeric Variables

SAS (Base SAS)+

data convert;
    input id $ char_num $ num_val;
    datalines;
1  100    55
2  200    60
3  ABC    65
4  300    .
5  .      75
;
run;

data convert2;
    set convert;

    * Convert character to numeric;
    char_to_num = input(char_num, best.);

    * Convert numeric to character;
    num_to_char = put(num_val, best.);
run;

SAS splits the two conversion directions across two functions: INPUT() goes character-to-numeric, PUT() goes numeric-to-character.
char_to_num = input(char_num, best.); reads the character value through a numeric informat and returns a numeric result.
The best. informat picks the most appropriate numeric reading automatically — good enough for most general-purpose conversions.
Invalid strings like "ABC" cannot be parsed as a number, so SAS returns missing and writes a note to the log.
num_to_char = put(num_val, best.); writes the numeric value through a format and returns a character string.
The best. format gives SAS latitude over width and decimal placement; use something tighter like 8.2 when the report needs a fixed layout.
We run and confirm convert2: the ABC row should have char_to_num = ., and the missing-num_val row should show an empty num_to_char.

R (tidyverse)+

convert <- tribble(
~id, ~char_num, ~num_val,
"1", "100", 55,
"2", "200", 60,
"3", "ABC", 65,
"4", "300", NA,
"5", NA, 75 )

 
convert2 <- convert %>%
   mutate( char_to_num = as.numeric(char_num),
     num_to_char = as.character(num_val)
)

In tidyverse, the two base-R conversion functions do the whole job — no stringr or forcats needed for this simple case.
as.numeric(char_num) reads each character value as a number; values that cannot be parsed become NA with a warning.
as.character(num_val) writes each numeric value as a string — NA stays NA.
Both calls are wrapped in a single mutate(), so the result keeps the original columns plus the two derived ones.
For stricter parsing, readr::parse_number() is an alternative that handles thousand separators, currency symbols, and similar dressings.
We run and confirm convert2: the ABC row has char_to_num = NA and the NA-num_val row has num_to_char = NA.

Happy learning!

Search lessons

Introduction

Convert Between Character and Numeric Variables