Partenaires

CNRS
Logo tutelle
Logo tutelle
Logo tutelle


Rechercher

Sur ce site

Sur le Web du CNRS


Accueil du site > Séminaires > Probabilités Statistiques et réseaux de neurones > Some experiences with Simulated Annealing in the solution of Statistical problems

Vendredi 2 février 2007 à 12h00

Some experiences with Simulated Annealing in the solution of Statistical problems

Carlos Bouza (Université La Havane, Cuba), bouza@matcom.uh.cu

Résumé : We present some experiences in the solution of different optimization problems that arise in Statistics. They are : · Optimum Multivariate stratification for maximizing the accuracy of estimation in sample surveys.Simulated annealing is used for constructing strata that minimize a variance function and the different approaches used in this study are presented (El. modelo de regresión lineal : algunas soluciones para su ajuste. IO, 20, & [1999]. 115-40., ; Optimum allocation and weighting in stratified sampling using Stochastic Programming. RT 96-03 CESMA-USB.Venezuela. ;. Model selection of the convex combination of LS and LAV : a simulated anneeling approach. RT.96-09 CESMA-USB.Venezuela ; Multiple linear regression curve fitting : a quadratic programming solution.[1995] In Approximation & Optimization” P. Lang Verlag, Frankfurt. 1-3, & ?! ; Interaction between Optimization and Statistics : Regression equation fitting and estimating the approximation error in Stochastic Programming. (2002), Proceedings of the Applied Mathematics Summer School, Humboldt University Berlin) · The study of inventory problems under random demandsSome applications of heuristics for solving the determination of an optimal inventory policy when random demands are present. The approach suggested has been developed in a series of papers (Bounds of the expected appoximation error in optimal inventory policies. [1998], 3rd. International Conf. on Approximation and Optimization in the Caribbean. &%+, 47-54. México ; Investigation of Burn-in-time problems with unknown failure time distribution . [2001] : J. of Statistics and Management Sc.37, 1-7. India.$ ;. A Study of the Optimum Lot Size and the Newsboy problem under Random Demands. (2003) Economic Analysis Working Papers, 3. Spain. ; Convergence of estimated optimal inventory levels in models with probabilistic demands. (2003), YUJOR 13, 217-227. Serbia Montenegro) · The solution of robust regression fitting and variable selectionSimulated annealing is used for solving problems related with the use of robust alternatives for LS regression. It allows to select variables without relying in the normality of the errors. The use of statistical models for studying the behavior of heuristics is also considered. Different algorithms and computational experiences are presented.

Dans la même rubrique :