Model regresi tersensor kiri pada nilai amatan nol

Abstract

Research aims at estimating censored regression model and ordinary regression model on 500 artificial observations containing zero value of dependent variables have been carried out. Criteria of best model were used according to properties of unbiasedness and minimum variance estimator. Results show that parameter estimate of censored regression model is more accurate than ordinary regression model. Censored regression model also yield smaller MSE (Mean Square Error) than ordinary regression model indicating that censored regression model is better than ordinary regression model.