Variable transformations


One of the advantages of programming in the SAS/IML language is its ability to transform data vectors with a single statement. For example, in data analysis, the log and square-root functions are often used to transform data so that the transformed data have approximate normality.

The following SAS/IML statements create a data vector, x, and apply several common transformations to the data:

proc iml;
x = {1,2,4,5,10,100};
y = log(x);   /** natural log **/
z = log10(x); /** log base 10 **/
u = sqrt(x);  /** square root **/

Notice that you do not need to write a DO loop in order to compute a vector of quantities: the LOG, LOG10, and SQRT functions transform every element of the vector x.

There are many other useful data transformations. In fact, entire books have been written about ways to transform data. The following statements show a few other common transformations:

v = 1 + 2 * x - 3 * x##2; /** polynomial transform **/
s = (x - mean(x)) / sqrt(var(x)); /** standardize **/
q = y + z/2 - 5 * u; /** linear combination **/

In the preceding statements, the MEAN and VAR functions are used. These functions are available in SAS/IML 9.22. If you are using an earlier version of SAS/IML, you can use the subscript reduction operator (:) (for example, (x-x[:])) and call the VAR module from my book.

The important point to remember is this: you do not need to write a loop in order to transform a data vector. Just write down the equation, and let the SAS/IML language do the rest.


About Author

Rick Wicklin

Distinguished Researcher in Computational Statistics

Rick Wicklin, PhD, is a distinguished researcher in computational statistics at SAS and is a principal developer of SAS/IML software. His areas of expertise include computational statistics, simulation, statistical graphics, and modern methods in statistical data analysis. Rick is author of the books Statistical Programming with SAS/IML Software and Simulating Data with SAS.

Leave A Reply

Back to Top