Answers are grounded in this paper and cite their sources.
An extended debate of the alleged virtues of LP as a substitute method for regression analysis by least-squares appeared in The Journal of Industrial Engineering. The motivation for this substitution is the ability of LP to restrain unwanted negative coefficients. After recounting the histories of these two methods and the Journal controversy, this article compares their use on a simple problem. Using n-dimensional space illustrations, the conditions which produce the “unexpected negative” phenomena are then set forth. A general theorem is presented. Two classes of inter-correlation are distinguished where one is generally helpful and the other generally harmful. Within this framework, the dangerous conditions of which sign-reversals are symptomatic are discussed with the conclusion that statistical problems should be treated with statistical analysis techniques.
No takes yet. Share an insight, caveat, or question.
Douglas C. Crocker (1969) studied this question.
Synapse has enriched 10 related papers. Consider them for comparative context: