SAS/ACCESS 9.2 for Relational Databases: Reference, Fourth Edition

Overview of SAS/ACCESS Interface to Relational Databases 4 Examples of Problems That Result in Numeric Imprecision 7Table 1.2Creating SAS ViewsFeature You Use to Access DBMS DataSAS/ACCESS LIBNAME statementSQL Pass-Through FacilitySAS View Technology You Can UseSQL view or DATA step view of the DBMS tableSQL view with CONNECTION TO componentChoosing Your Degree of Numeric PrecisionFactors That Can Cause Calculation DifferencesDifferent factors affect numeric precision. This issue is common for many people,including SAS users. Though computers and software can help, you are limited in howprecisely you can calculate, compare, and represent data. Therefore, only those peoplewho generate and use data can determine the exact degree of precision that suits theirenterprise needs.As you decide the degree of precision that you want, you need to consider that thesesystem factors can cause calculation differences:3 hardware limitations3 differences among operating systems3 different software or different versions of the same software3 different database management systems (DBMSs)These factors can also cause differences:3 the use of finite number sets to represent infinite real numbers3 how numbers are stored, because storage sizes can varyYou also need to consider how conversions are performed—on, between, or across anyof these system or calculation factors.Examples of Problems That Result in Numeric ImprecisionDepending on the degree of precision that you want, calculating the value of r canresult in a tiny residual in a floating-point unit. When you compare the value of r to0.0, you might find that r≠0.0. The numbers are very close but not equal. This type ofdiscrepancy in results can stem from problems in representing, rounding, displaying,and selectively extracting data.Representing DataSome numbers can be represented exactly, but others cannot. As shown in thisexample, the number 10.25, which terminates in binary, can be represented exactly.data x;x=10.25;put x hex16.;run;The output from this DATA step is an exact number: 4024800000000000. However,the number 10.1 cannot be represented exactly, as this example shows.