Advances and Applications in Statistics
Volume 60, Issue 2, Pages 137 - 146
(February 2020) http://dx.doi.org/10.17654/AS060020137 |
|
A COMPARISON BETWEEN RIDGE REGRESSION AND LEAST ABSOLUTE VALUE USING SIMULATION TECHNIQUE
M. M. T. Al-Kassab and M. Almousa
|
Abstract: A basic assumptions concerned with general linear regression model is that there is no correlation between the explanatory variables. When this assumption is not satisfied, the least squares estimators have large variances and covariances and become unstable and may have wrong sign. So, we resort to biased regression method, which stabilize the parameter estimates. Ridge regression and least absolute value methods are two of the most popular biased regression method. In this research, we use a simulated data, using Monte Carlo method, to estimate the regression coefficients by ridge regression and least absolute value methods. We generate a data set that consists of twenty variables and five different level of correlation α2 = (0.35, 0.51, 0.67, 0.84, 0.99). A comparison between Ridge regression and least absolute value methods is made in the sense of mean squares error criterion and find that the least absolute value is the best. |
Keywords and phrases: ridge regression, mean square error, least absolute deviation.
|
|
Number of Downloads: 275 | Number of Views: 1032 |
|