5.2 Numeric & Mathematical Functions

5.2 Numeric & Mathematical Functions

Numeric processing is at the heart of SAS programming. Whether performing financial calculations, data transformations, or descriptive statistical analysis, SAS programmers rely on built-in numeric functions. A critical requirement for the SAS Base Certification exam is knowing how SAS numeric functions handle missing data, rounding precision, integer truncation, mathematical transformations, and variable list arguments.

---

1. Rounding and Truncation Functions

SAS provides several functions to round or truncate floating-point numbers into integers or specified decimal increments.

The ROUND Function

The ROUND function rounds a numeric value to the nearest multiple of a specified round-off unit.

Syntax: ROUND(argument, <round-off-unit>)

• If round-off-unit is omitted, SAS rounds to the nearest integer (default unit = 1).
• If the argument is halfway between two multiples, SAS rounds up to the larger value.

data work.round_demo;
    val1 = round(12.3456, 0.01);  /* 12.35 (rounds to nearest hundredth) */
    val2 = round(12.3456, 0.1);   /* 12.3  (rounds to nearest tenth)     */
    val3 = round(12.3456);        /* 12    (rounds to nearest integer)   */
    val4 = round(148.50, 10);     /* 150   (rounds to nearest ten)       */
run;

Comparing INT, FLOOR, and CEIL

Understanding the differences between integer truncation functions is frequently tested on the SAS certification exam, especially when operating on negative numbers.

data work.trunc_demo;
    pos = 5.8;
    neg = -5.8;
    
    int_pos   = int(pos);    /*  5 */
    floor_pos = floor(pos);  /*  5 */
    ceil_pos  = ceil(pos);   /*  6 */
    
    int_neg   = int(neg);    /* -5 (truncates decimal) */
    floor_neg = floor(neg);  /* -6 (rounds down away from zero) */
    ceil_neg  = ceil(neg);   /* -5 (rounds up toward zero) */
run;

---

2. Advanced Mathematical Utility Functions (MOD, SQRT, ABS, LOG)

In addition to rounding, Base SAS includes mathematical functions for algebra, remainder arithmetic, and absolute value calculations.

The MOD Function (Remainder Division)

MOD(n, d) returns the remainder when n is divided by d.

data work.mod_demo;
    rem1 = mod(10, 3); /* Returns 1 (10 / 3 = 3 remainder 1) */
    rem2 = mod(14, 5); /* Returns 4 */
run;

/* Exam Application: splitting even and odd observation numbers.
   Every data set written by OUTPUT must be named on the DATA statement. */
data work.even_rows work.odd_rows;
    set work.transactions;
    if mod(_n_, 2) = 0 then output work.even_rows;
    else output work.odd_rows;
run;

Additional Mathematical Functions

• ABS(x): Returns the absolute (positive) value of x (e.g., ABS(-15) = 15).
• SQRT(x): Returns the square root of x. If x is negative, SAS returns a missing value (.) and prints an invalid argument note to the log.
• LOG(x) / EXP(x): Returns the natural logarithm or exponential power ($e^x$) of x.
• SIGN(x): Returns -1 if x < 0, 0 if x = 0, and 1 if x > 0.

Order-Statistic and Random Functions Named in the Exam Content Guide

The official A00-231 content guide explicitly names SMALLEST, LARGEST, and RAND alongside SUM, MEAN, ROUND, and INT, so expect at least one item drawn from them.

• SMALLEST(k, value-1, ..., value-n): Returns the k-th smallest non-missing value from the argument list. SMALLEST(1, 40, ., 12, 75) returns 12; SMALLEST(2, 40, ., 12, 75) returns 40.
• LARGEST(k, value-1, ..., value-n): Returns the k-th largest non-missing value. LARGEST(1, 40, ., 12, 75) returns 75; LARGEST(2, ...) returns 40.
• RAND('distribution', <parameters>): Returns a random number from the named distribution. RAND('UNIFORM') returns a value in the interval (0, 1); RAND('NORMAL', 100, 15) returns a normal draw with mean 100 and standard deviation 15; RAND('INTEGER', 1, 6) simulates a die roll.

data work.order_stats;
    a = 40; b = .; c = 12; d = 75;

    low1  = smallest(1, a, b, c, d);  /* 12 - missing values are skipped */
    low2  = smallest(2, a, b, c, d);  /* 40 */
    high1 = largest(1, a, b, c, d);   /* 75 */
    spread = high1 - low1;            /* 63 */
run;

/* RAND requires a seed for reproducible streams */
data work.simulated;
    call streaminit(20260807);   /* fixes the random stream */
    do trial = 1 to 5;
        u = rand('UNIFORM');
        die = rand('INTEGER', 1, 6);
        output;
    end;
run;

Exam Tip: SMALLEST(k, ...) and LARGEST(k, ...) take the rank first and the values afterwards, and both ignore missing values. RAND needs CALL STREAMINIT(seed); before the first call if the results must be reproducible; without it SAS seeds from the system clock.

---

3. Descriptive Statistics Functions & Missing Values

SAS DATA step functions can calculate summary statistics across variables within a single observation (row-wise processing).

Common statistical functions include:

• SUM(arg1, arg2, ...): Calculates the total of non-missing values.
• MEAN(arg1, arg2, ...): Calculates the arithmetic average of non-missing values.
• MIN(arg1, arg2, ...) / MAX(arg1, arg2, ...): Identifies the minimum or maximum non-missing value.
• N(arg1, arg2, ...): Counts the number of non-missing numeric values.
• NMISS(arg1, arg2, ...): Counts the number of missing numeric values.
• STD(arg1, arg2, ...) / VAR(arg1, arg2, ...): Computes the standard deviation or variance across row arguments.

Critical Difference: SUM Function vs. Addition Operator (+)

One of the most important concepts in SAS programming is the handling of missing values (.) when performing arithmetic calculations.

data work.missing_comparison;
    x = 10;
    y = .;
    z = 30;
    
    /* Using Addition Operator */
    total_op = x + y + z;   /* Result: . (Missing!) */
    
    /* Using SUM Function */
    total_fn = sum(x, y, z); /* Result: 40 (Ignores missing value) */
run;

Exam Rule: If any argument in an expression using the + operator is missing, SAS evaluates the entire expression as a missing value (.). In contrast, the SUM function ignores missing values and calculates the sum of all remaining valid numeric arguments. If all arguments passed to SUM are missing, SUM returns a missing value.

---

4. Using Variable Lists with the OF Keyword

When passing a list or range of variables to a SAS function, you must include the keyword OF before the variable list.

Variable List Syntax Types:

• Numbered Range: sum(of qtr1-qtr4) (expands to qtr1, qtr2, qtr3, qtr4)
• Name Range: mean(of height--weight) (includes all variables between height and weight in dataset order)
• Name Prefix: max(of sales:) (includes all variables starting with the prefix sales)
• All Numeric: sum(of _numeric_) (includes all numeric variables in the dataset)

data work.variable_lists;
    qtr1 = 100; qtr2 = 150; qtr3 = .; qtr4 = 200;
    
    /* Correct syntax using OF keyword */
    total_sales = sum(of qtr1-qtr4);  /* Result: 450 */
    avg_sales   = mean(of qtr1-qtr4); /* Result: 150 (450 / 3 valid values) */
    
    /* INCORRECT SYNTAX (Common Exam Trap!) */
    /* diff = sum(qtr1 - qtr4); */
    /* Without OF, SAS computes qtr1 minus qtr4 (100 - 200 = -100) and passes -100 to SUM! */
run;

Critical Exam Pitfall: Omitting the OF keyword in sum(qtr1-qtr4) causes SAS to treat - as a subtraction operator rather than a range hyphen, yielding sum(100 - 200) = -100 instead of summing four variables!