Comparison of Cochrane-Orcutt and Hildreth-Lu Methods to Overcome Autocorrelation in Time Series Regression (Case Study of Gorontalo Province HDI 2010-2021)

Abstract

Time series data is data that is prone to autocorrelation. Autocorrelation is a violation of assumptions in Ordinary Least Square regression. The presence of autocorrelation can make parameter estimates, not BLUE (Best, Linear, Unbiased Estimator). Several methods to overcome autocorrelation include Cochrane-Orcutt and Hildreth-Lu methods. Therefore, this study aimed to compare the Cochrane-Orcutt and Hildreth-Lu methods to deal with autocorrelation in the time series regression of the Gorontalo Human Development Index case in 2010 2021. We used HDI data for Gorontalo Province from 2010-to 2021, taken from the BPS-Statistics Indonesia Gorontalo Province. The method we used was Cochrane-Orcutt and Hildreth-Lu in the case of regression using Ordinary Least Squares (OLS) parameter estimation. The results obtained are that the Cochrane-Orcutt and Hildreth-Lu could overcome autocorrelation. The results of the Durbin Watson test after using both methods show no autocorrelation. However, the Hildreth-Lu method resulted in a lower Root Mean Square Error (RMSE) of 0.147 compared to the RMSE of the OLS model of 0.165 and the RMSE of the Cochrane-Orcutt model of 0.196. Therefore, the Hildreth-Lu method was the best method to overcame autocorrelation in this case.