The HP4SCORE Procedure
PROC HP4SCORE Contrasted with Other SAS Procedures
No other SAS procedure scores a forest model. SAS high-performance procedures create DATA step programs that incorporate the scoring logic in the model. The decision tree procedures in SAS Enterprise Miner (the ARBOR and HPSPLIT procedures) also output the score code directly. Because the forest model is an ensemble of hundreds of decision trees and because each decision tree can contain hundreds or thousands of decision rules for scoring, the DATA step program can become extremely large. In addition to slow scoring performance, a very large DATA step program also poses other challenges. PROC HP4SCORE overcomes these difficulties by reading the binary model file and scoring the observations directly.
The HP4SCORE procedure takes full advantage of concurrent threads and distributed data. For general contrasts between SAS high-performance analytical procedures and other SAS procedures, see the section Output Data Sets.