Article ID | Journal | Published Year | Pages | File Type |
---|---|---|---|---|
416328 | Computational Statistics & Data Analysis | 2015 | 14 Pages |
Abstract
An exact algorithm for computing the estimates of regression coefficients given by the least trimmed squares method is presented. The algorithm works under very weak assumptions and has polynomial complexity. Simulations show that in the case of two or three explanatory variables, the presented algorithm is often faster than the exact algorithms based on a branch-and-bound strategy whose complexity is not known. The idea behind the algorithm is based on a theoretical analysis of the respective objective function, which is also given.
Keywords
Related Topics
Physical Sciences and Engineering
Computer Science
Computational Theory and Mathematics
Authors
Karel Klouda,