Jim Simon is a principal instructor and course developer for SAS Education. Jim has a bachelor's degree from UCLA and a master's degree from California State University at Northridge. Prior to joining the SAS Irvine office in 1988, Jim was an instructor at Ventura College and a SAS programmer at The Medstat Group in Santa Barbara. Jim's areas of specialization include the DATA step, application development, web enablement, and the SAS macro language. A native of Southern California, Jim enjoys anything in the warm California sun. On weekends, Jim loves jumping in his Corvette, turning up the stereo, and cruising Pacific Coast Highway, top down, South to Laguna Beach or North to his old home town, Santa Barbara.
In a previous blog, Random Sampling: What's Efficient?, I discussed the efficiency of various techniques for selecting a simple random sample from a large SAS dataset. PROC SURVEYSELECT easily does the job: proc surveyselect data=large out=sample method=srs /* simple random sample */ rate=.01; /* 1% sample rate */ run; Note:
Suppose you wish to select a random sample from a large SAS dataset. No problem. The PROC SURVEYSELECT step below randomly selects a 2 percent sample: proc surveyselect data=large out=sample method=srs /* simple random sample */ n=1000000; /* sample size */ run; Do you have a SAS/STAT license? If not,
This post is the third and final in a series that illustrates three different solutions to "flattening" hierarchical data. Don't forget to catch up with Part 1 and Part 2. Solution 2, from my previous post, created one observation per header record, with detail data in a wide format, like
