JACKKNIFED RIDGE REGRESSION ESTIMATOR UNTUK MODEL LINEAR DENGAN AUTOKORELASI PADA ERROR; JACKKNIFED RIDGE REGRESSION ESTIMATOR IN THE LINEAR MODEL WITH CORRELATED ERRORS

Sari, Naomi Ratna, Subanar

2015 | Skripsi | FMIPA UGM

Regression analysis is a statistical analysis that used to perform model relationship between dependent variable and independen variables. One of the assumption in classical regression analysis is there is no multicollinearity and no autocorrelated errors. If there is multicollinearity and autocorrelated errors in the regression model, it could cause the result of the model that using the method of Least Squares Estimator becomes invalid. This paper will discuss about estimating ridge regression parameters using jackknifed ridge regression estimator in the linear model with autocorrelated errors developed by Özkale (2008). This method is extension of ridge regression estimator pioneered by Hoerl and Kennard (1970). This paper case study is using the amount of money circulating in Indonesia and the factors that affecting it from July 2007 till November 2014. The conclusion is the jackknifed ridge regression estimator gives smallerMSE than the generalized least square estimator and smaller bias than the ridge regression estimator.

Kata Kunci : multicollinearity; correlated errors; least square regression; generalized least square; ridge regression; jackknifed technique; MSE.