Application of TSA to uncapacitated P-median problem

In this thesis, application of TSA on discrete uncapacitated P-median problem is examined. Purpose of this thesis is to prove applicability of TSA on discrete P-median problem. Algorithm is tested on OR-Library data sets. P-median problem is considered as location network problem and it is also applicable to real life problems. Purpose of the P-median problem studied in this thesis is selecting location of given number of medians (centers) among the other nodes in order to minimize total cost, while satisfying all demand cost. The significant constraint which required to be taken in consideration is one demand point can be served just by one median. When it comes to TSA, it can be classified as nature-inspired, population and evolutionary based algorithm. During the application TSA, in order to have more accurate evaluation about TSA, algorithm have been run for five times for each data set. In computational results, CPU time and average of found minimum results are taken into account to evaluate performance of algorithm. For the small and medium sized data sets, results are satisfactory. Also, it can be said that number of medians are effective on deviation from optimal solution and CPU time end iterations. Meanwhile, total number of nodes is said to be effective on CPU time but not as much as effective compared to number of medians.