Principal components regression estimators under biased stochastic linear restrictions

OpenAlex İngilizce

We propose new estimators to combat multicollinearity in linear regression model under biased stochastic linear restrictions. The new estimators are constructed by combining weighted mixed estimator and principal components regression estimator. Furthermore, necessary and sufficient conditions for the superiority of the new estimators over ordinary least squares estimator, principal components regression estimator and ridge estimator are derived in the sense of the mean square error criterion. Finally, a numerical example and a broad Monte Carlo simulation experiment are conducted to demonstrate the validation of theoretical results.