Open AccessEstimasi Parameter Model Autoregressive dengan Metode Yule Walker, Least Square, dan Maximum Likelihood (Studi Kasus Data ROA BPRS di Indonesia)
Author(s) -
Maulida Nurhidayati
Publication year - 2021
Publication title -
quadratic
Language(s) - English
Resource type - Journals
eISSN - 2776-9003
pISSN - 2776-8201
DOI - 10.14421/quadratic.2021.011-01
Subject(s) - mathematics , autoregressive model , statistics , maximum likelihood , maximum likelihood sequence estimation , univariate , autoregressive integrated moving average , estimation theory , restricted maximum likelihood , time series , multivariate statistics
The Autoregressive model is a time series univariate model for stationary models. In estimating parameters on this model can be done by several methods, namely yule-walker method, Least Square, and Maximum Likelihood. Each method has a different principle for estimating model parameters so that the results obtained will also be different. Based on this, in this study, the AR(1) model parameter estimation was estimated by generating data simulated 1000 times to see the performance of Yule-Walker, Least Square, and Maximum Likelihood methods. In addition, the comparison of these three methods is also done on ROA BPRS data that follows the AR(1) model. The results showed that the Maximum Likelihood method was able to provide mode results and comparison of the most suitable estimation results for simulation data and produce the smallest MAE values in the data in sample and MAPE, MSE, and MAE the smallest in the out sample data. These results show that the Maximum Likelihood method is the best method for modeling data that follows the AR(1) model.