A review of the status and use of validation procedures for heuristics used in forest planning

Pete Bettinger, John Sessions, Kevin Boston

Abstract


While there exist clear methods for validating and ensuring the quality of solutions generated by forest planning heuristic techniques, the use of these methods in the literature varies from one situation to the next. Based on our experience developing and using heuristic forest planning techniques, we describe six levels of heuristic validation that are currently in use, ranging from no validation (Level 1) on one end of the spectrum, to the comparison of heuristic technique solutions with an exact solution obtained using mathematical programming methods (Level 6) on the other end. The reasons why authors may choose or reviewers may require levels of validation are proposed. We do not believe that all research papers should be subjected to the highest level of validation, but suggest that authors of papers on forest planning techniques and reviewers associated with peer-reviewed journals try to place the level of validation within the larger scientific context, then determine an appropriate level of validation. Admittedly, this is problematic for review decisions, given the fact that reviewers may differ in opinion of what is appropriate. Four brief cases are provided to help one think through these issues. Ultimately, we hope that this discussion will lead to a reasoned approach for the use of validation processes in conjunction with the presentation of heuristic techniques, rather than the current ad-hoc process that, on one hand, relies on the valuable and careful thoughts of the reviewers, yet on the other hand, may be uneven in application. MCFNS-1:26-37.


Keywords


combinatorial optimization, extreme value theory, linear programming, mixed integer programming, operations research

Full Text:

PDF PS

Refbacks

  • There are currently no refbacks.


   

© 2008 Mathematical and Computational Forestry & Natural-Resource Sciences