As a result of exiting high-dimensional ill-posed results,
more attention has been provided to regularisation methods in the last two
decades. To obtain more meaningful predictors, a smaller subset of large
numbers of predictors is required. A new way of incorporating the penalised
concept in regularised regression is proposed in this paper. The proposed
penalty is based on using the variances of the regression parameters from the
least square estimator. Some penalised estimators such as ridge, lasso, and elastic
net, which are used to resolve both the problem of multicollinearity and to
select variables, are added to the proposed system. Using the average mean
squared error criterion (AMSE), good results are achieved. For simulated
details, the best results in the simulated data are also seen in real
data. The type of the resulting
estimators' lower average prediction errors (APE).
Author (s) Details
Magda M. M. Haggag
Department of Statistics, Mathematics and Insurance, Faculty of Commerce,
Damanhour University, Egypt.
View Book :- https://bp.bookpi.org/index.php/bpi/catalog/book/271
No comments:
Post a Comment