Pendugaan parameter model dinamik dengan metode Median Absolute Deviation (MAD) dan Huber M-Estimation

Abstract

Parameter estimation is commonly applied to regression models. However, the estimation of the dynamic models have not been developed. Least Square Method is the most common method used in parameter estimation. However, this method is not appropriate to be used if data contains some outliers. Robust method is a method that can overcome the weakness of the Least Square method. Median Absolute Deviation and M-Estimation are some suitable methods used when the data contains outliers and far outliers. Both of these robust methods are quite good in parameter estimation for the data with or without outliers. Based on Symmetrical Mean Absolute Percentage Error (SMAPE) and boxplot, MEstimation method has better accuracy in parameter estimation than the Median Absolute Deviation method. In this manuscript, parameter estimation is applied to Gompertz and SZR (Susceptible, Zombie, Removed) model using hypothetical data.