Least squared approach to the traveling salesman problem

Abstract

Scope of Study: This study presents an alternative method to solving multipoint distribution problems. The method provides for potential improved route selection for the distribution problem known as the traveling salesman. The method comprises determining the number of transportation units necessary using the convex hull area, a minimum sided polygon. The individual route selection is found by first performing a cluster analysis to divide the stations into first level groups. Generation of a regression line for each group and projection of each station onto this regression line establishes the route sequence. The distribution point is integrated into the route using the shortest distance path length. The affects of various parameters are also discussed.

Findings and Conclusions: The solution to a randomly generated distribution problem showed a near equivalent solution as compared to another method known as lockset. The method resulted in different solution path lengths. This method provides an alternative approach to solving distribution problems involving a common distribution origin. It also can present potential savings in distance and thus costs by its application. The method provides a viable alternative to other methods.