BUILDING LINEAR PREDICTIVE MODELS FOR URBAN PLANNING

BECAUSE LINEAR MODELS ARE BEING USED MORE OFTEN FOR PREDICTIVE PURPOSES, IT IS ESSENTIAL THAT THE UNDERLYING ASSUMPTIONS AND LIMITATIONS OF THESE MODELS ARE UNDERSTOOD. A MODEL CAN CONTAIN TWO BASIC TYPES OF ERROR; MEASUREMENT AND SPECIFICATION ERROR. MEASUREMENT ERRORS ARE INCURRED IN DATA COLLECTION, SCALING, AND SAMPLING. SPECIFICATION ERRORS RANGE FROM THE INCLUSION OF INTER-CORRELATED VARIABLES AND NON-LINEAR RELATIONSHIPS TO THE FAILURE TO CORRECTLY EVALUATE THE MODEL. THESE ERRORS WHICH ACCUMULATE AS MODELS ARE MERGED INTO LARGER MODEL SYSTEMS SO THAT ACCURATE PREDICTION BECOMES INCREASELY DIFFICULT. AN EXAMPLE IS PRESENTED FOR STRUCTURING A MODEL WHICH CONCENTRATES ON CHECKS AND NON-ADDITIVE ERROR DESIGNS, SIMILAR TO THE SAN FRANCISCO HOUSING MARKET MODEL. MODEL EVALUATION IS AN IMPORTANT ASPECT OF MODEL BUILDING AND SHOULD INCLUDE ESTIMATION OF THE SENSITIVITY AND STABILITY OF PARAMETERS. DESPITE THEIR LIMITATIONS LINEAR MODELS ARE USEFUL IF DESIGNED AND INTERPRETED CAREFULLY. /AUTHOR/

• Availability:
  • Find a library where document is available. Order URL: http://worldcat.org/issn/00343404
• Supplemental Notes:
  • Vol 2, pp 139-143
• Authors:
  • Colenutt, R J
• Publication Date: 0

Media Info

• Serial:
  • Regional Studies
  • Publisher: Taylor & Francis
  • ISSN: 0034-3404
  • EISSN: 1360-0591
  • Serial URL: http://www.tandfonline.com/loi/cres20

Subject/Index Terms

• TRT Terms: Design; Linear equations; Mathematical models; Traffic simulation
• Subject Areas: Design; Highways; Operations and Traffic Management;

Filing Info

• Accession Number: 00227265
• Record Type: Publication
• Source Agency: Traffic Systems Reviews & Abstracts
• Files: TRIS
• Created Date: Jul 27 1970 12:00AM