White Test for the Least Weighted Squares
Author: | Bc. Věra Bludská |
---|---|
Year: | 2011 - summer |
Leaders: | prof. RNDr. Jan Ámos Víšek CSc. |
Consultants: | |
Work type: | Bachelors |
Language: | English |
Pages: | 48 |
Awards and prizes: | B.A. with distinction from the Dean of the Faculty of Social Sciences for an extraordinarily good bachelors diploma thesis. |
Link: | |
Abstract: | The Least Weighted Squares (LWS) is a robust method for computing coefficients in linear regression models. An inherent problem of LWS is the complexity of its estimator and, consequently, the lack of an analytical solution or fast exact algorithms for its evaluation. To remedy this situation a novel exact algorithm running in polynomial time has been proposed. The algorithm implemented in MATLAB programming language has been employed for testing computationally more efficient non-exact LWS methods. In addition to many potential uses of LWS in robust econometrics (e.g. outlier diagnostics) the method has been applied to the problem of regression estimation in the presence of heteroscedasticity. It has been demonstrated that the combined use of the LWS estimator and White's test for heteroscedasticity significantly improves the efficiency of the robust regression estimation. |
Downloadable: | Bachelor Thesis of Bludská |