5.5 Explicit & Implicit Data Type Conversions (INPUT & PUT)

5.5 Explicit & Implicit Data Type Conversions (INPUT & PUT)

SAS supports exactly two data types: numeric and character. Frequently, incoming datasets contain numbers stored in character variables (e.g., "$1,250.50" or "20260101") or numeric values that must be formatted as character strings for concatenation or reporting. Converting data between character and numeric types—both explicitly and implicitly—is one of the most frequently examined domains on the SAS Certified Specialist Base Programming exam.

---

1. Explicit Character-to-Numeric Conversion: The INPUT Function

The INPUT function converts character values to numeric or character values based on a specified INFORMAT.

$\text{Syntax: } \text{target\_variable} = \text{INPUT}(\text{source\_variable}, \text{informat.});$

• source_variable: The character variable or string to be converted.
• informat.: The SAS informat that instructs SAS how to read and interpret the source string.

data work.convert_char_to_num;
    char_price  = "$1,250.75";
    char_date   = "10/31/2026";
    
    /* Convert formatted character currency to numeric */
    num_price = input(char_price, dollar10.2);
    
    /* Convert character date string to SAS date integer */
    num_date  = input(char_date, mmddyy10.);
    
    format num_date date9. num_price dollar10.2;
run;

[!IMPORTANT] Memory Rule: The INPUT function requires an INFORMAT (with a dot). The data type of target_variable is determined by the informat used. If the informat is numeric (such as 8. or COMMA9.), target_variable becomes numeric.

---

2. Explicit Numeric-to-Character Conversion: The PUT Function

The PUT function converts numeric or character values into character values formatted according to a specified FORMAT.

$\text{Syntax: } \text{target\_variable} = \text{PUT}(\text{source\_variable}, \text{format.});$

• source_variable: The numeric variable or expression to be converted.
• format.: The SAS format that instructs SAS how to write out the character string.

data work.convert_num_to_char;
    num_code  = 4096;
    sas_date  = 24107; /* Represents 01JAN2026 */
    
    /* Convert numeric integer to 6-character string padded with leading zeros */
    char_code = put(num_code, z6.);     /* Result: '004096' */
    
    /* Convert SAS date integer to character string formatted as YYYY-MM-DD */
    char_date = put(sas_date, yymmdd10.); /* Result: '2026-01-01' */
run;

[!IMPORTANT] The PUT function always returns a character string. Therefore, target_variable is always created as a character variable.

---

3. Variable Re-assignment Rules & Type Conflicts

In a SAS DATA step, a variable's data type is established upon its first reference during compilation and cannot be changed during execution. A widespread misconception is that re-assigning a converted value to the same variable name raises an error. It does not — SAS silently converts the value back, which is more dangerous than an error because the program keeps running with corrupted data.

/* WRONG - but NOT an error. SAS silently converts the number back to text! */
data work.error_example;
    salary = "50000";           /* SALARY is CHARACTER, length 5     */
    salary = input(salary, 8.); /* INPUT returns numeric 50000 ...   */
run;                            /* ... then SAS converts it back to  */
                                /* character using BEST5.            */

SAS Log Output: NOTE: Numeric values have been converted to character values at the places given by: (Line):(Column).

SALARY remains a character variable. Per the SAS 9.4 language reference, when a numeric result is assigned to a character variable SAS applies the BESTw. format, where w is the length of the character variable (maximum 32). Because SALARY is 5 bytes, BEST5. reproduces '50000'. Had SALARY been declared $3, the number would not fit and SAS would fill the variable with asterisks ('***') and note invalid character data.

Correct Pattern for In-Place Conversion:

To replace a character variable with a numeric variable of the same name, use dataset options (RENAME= and DROP=) during compilation:

data work.correct_reassign (drop=char_salary);
    set work.raw_data (rename=(salary=char_salary));
    salary = input(char_salary, 8.); /* New numeric variable SALARY */
run;

---

4. Implicit (Automatic) Data Type Conversions

When a variable of one data type is used in a context that requires the other data type, SAS attempts an implicit type conversion automatically.

Character-to-Numeric Implicit Conversion

Implicit character-to-numeric conversion occurs when a character variable is used in a numeric context, such as:

• Arithmetic operators (+, -, *, /)
• Logical comparison with numeric constants
• Assignment to a numeric variable

data work.implicit_num;
    char_val = "100";
    total = char_val + 50; /* Implicitly converts char_val to numeric 100 */
run;

SAS Log Output: NOTE: Character values have been converted to numeric values at the places given by: Line 3 Column 13.

What Happens If Implicit Conversion Fails?

If char_val contains non-numeric characters (e.g., "100ABC" or "$50"), SAS cannot perform the implicit conversion:

1. SAS sets the target variable to missing (.).
2. SAS sets the automatic variable _ERROR_ = 1.
3. SAS writes a note to the log: NOTE: Invalid numeric data, '100ABC' , at line 3 column 13.

Numeric-to-Character Implicit Conversion

Implicit numeric-to-character conversion occurs when a numeric variable appears in a character context, such as:

• Character functions (e.g., SUBSTR, SCAN, CATX)
• Concatenation operator (||)
• Assignment to a character variable

When converting numeric to character implicitly, SAS applies the BEST12. format, right-aligning the result in a 12-byte string.

data work.implicit_char;
    num_id = 789;
    /* Implicitly converts num_id using BEST12. format */
    combo = "ID-" || num_id; 
run;

[!CAUTION] Because SAS uses BEST12. for implicit numeric-to-character conversion, the resulting string contains leading blanks (9 spaces followed by 789). Thus, combo becomes "ID- 789"! Perform an explicit conversion and strip the padding instead: combo = "ID-" || strip(put(num_id, best12.)); or simply combo = cats("ID-", num_id);. Note that strip() must wrap the whole PUT() call — put(num_id, strip(best.)) is invalid syntax, because the second argument of PUT must be a format, never a function call.

---

5. Summary Matrix: INPUT vs. PUT