AN ITERATIVE PROCEDURE FOR OUTLIER DETECTION IN GSTAR(1;1) MODEL

Outliers are observations that differ significantly from others can affect the estimation results in model and reduce estimator's accuracy. To deal with outliers is to remove data. However, sometimes important information contained outlier, so eliminating a misinterpretation. There two types of time series model, Innovative Outlier (IO) Additive (AO). In GSTAR spatial correlations also be detected. We introduce an iterative procedure for detecting model. The first step form without outlier factors. Furthermore, detection model's residuals. If detected, add factor into initial estimate parameters new residuals obtained process repeated by adding them until no As result, not removed or ignored but This paper presents case studies about Dengue Hemorrhagic Fever cases five locations West Kalimantan Province. These subject result this using detect based on residual provides better accuracy than regular (without model). It solved removing data factors way, critical id lost, accurate ore obtained.