Open AccessWeighting Temporary Change Outlier by Modified Huber Function with Monte Carlo Simulations
Author(s) -
Intan Martina Md Ghani,
H. Abdul Rahim
Publication year - 2020
Publication title -
journal of physics. conference series
Language(s) - English
Resource type - Journals
SCImago Journal Rank - 0.21
H-Index - 85
eISSN - 1742-6596
pISSN - 1742-6588
DOI - 10.1088/1742-6596/1529/5/052050
Subject(s) - outlier , weighting , weight function , monte carlo method , statistics , mathematics , function (biology) , series (stratigraphy) , mean squared error , econometrics , computer science , medicine , paleontology , evolutionary biology , biology , radiology
Robust method is a popular approach to dealing the existence of outliers in the data. Many researchers have applied Huber weight function. The aim of this paper is to evaluate the performance of the Huber weight function and the modification of the Huber weight function on the temporary change (TC) outliers. The data used in this paper were generated as ARMA(1,0)-GARCH(1,2) model via the Monte Carlo simulation. There are three major situations in this simulations: without weight (WW), with Huber weight (WH) and with a modified Huber weight (WMH). Three different TC contamination (0%, 10% and 20%) and three different time series length (100, 500 and 1000) were tested. The performance of the three situations was compared on the basis of AIC, SIC, HQIC, MAE, MSE and RMSE. The results of the numerical simulations show that the performance in the WMH situation is better than the WH situation in the presence of TC outliers.