← SASnRRound Numeric Values

Introduction

Rounding is a daily need in report layout and derived-variable building — to an integer, to a decimal place, up to the next integer, down to the previous one.
Four common asks: round to nearest integer, round to a decimal place, ceiling, and floor.
SAS offers round(value) for nearest integer, round(value, 0.01) for a decimal place, and ceil / floor for directional rounding.
R uses round() with an optional digits argument, plus ceiling() and floor().
A subtle difference: R's round() uses banker's rounding (round-half-to-even), whereas SAS rounds half up.
This lesson rounds a short vector that includes 1.5 and -4.75 so the half-case difference becomes visible.

Round Numeric Values

SAS (Base SAS)+

data round_demo;
    input value;
    datalines;
3.1416
2.718
9.99
1.5
-4.75
;
run;

data round_results;
    set round_demo;
    round_val = round(value);
    round_dec = round(value,0.01);
    ceil_val = ceil(value);
    floor_val = floor(value);
run;

In SAS, round(value) rounds to the nearest integer using the round-half-up rule.
round(value, 0.01) rounds to the nearest two decimal places; the second argument is the rounding unit, not a digit count.
ceil(value) returns the smallest integer greater than or equal to the value — always steps upward.
floor(value) returns the largest integer less than or equal to the value — always steps downward.
On 1.5, SAS round() returns 2; on -4.75 rounded to two decimals, it returns -4.75 (already aligned).
We run and confirm each derived column reflects the intended rounding rule; pay attention to the 1.5 row as the half-case test.

R (tidyverse)+

round_demo <- tribble( ~value, 
                     3.1416, 
                     2.718, 
                     9.99, 
                     1.5, 
                     -4.75 ) 

round_results <- round_demo %>% 
  mutate( 
    round_val = round(value), 
    round_dec = round(value,2), 
    ceil_val = ceiling(value), 
    floor_val = floor(value) 
    )

In R, round(value) rounds to the nearest integer, but uses banker's rounding (round-half-to-even).
round(value, 2) rounds to two decimal places; the second argument is a digit count, not a rounding unit — a small naming difference from SAS.
ceiling(value) matches SAS CEIL(); floor(value) matches SAS FLOOR().
For exact parity with SAS on half-cases, a custom helper round_sas <- function(x, n = 0) trunc(x * 10^n + sign(x) * 0.5) / 10^n replicates round-half-away-from-zero.
The tidyverse wrapper here is just mutate() with the base R functions — no extra packages needed for simple rounding.
We run and compare the round_val column across SAS and R: 1.5 rounds to 2 in both, but 2.5 would round to 2 in R and 3 in SAS.

Happy learning!

Search lessons

Introduction

Round Numeric Values