期刊名称：Majalah Iptek = IPTEK : The Journal for Technology and Science
印刷版ISSN：0853-4098
电子版ISSN：2088-2033
出版年度：2008
卷号：19
期号：4
DOI：10.12962/j20882033.v19i4.140
语种：English
出版社：IPTEK
摘要：Estimation of regression curve usually conducted using three methods; parametric method, non-parametric method, and semi-parametric method. Non-parametric method has several techniques, which are histogram, kernel, and spline. From various types of spline techniques, weighted parsial spline is developed to solve heterokedasticity problem, this is due to the inability of original partial spline model in handling the heterokedasticity problem. Different techniques are used in choosing the weighted criteria, one of the technique is Generalized Moving Average (GMA). Study about the amount of electricity power loss in PT. PLN East Java Province, North Surabaya Region, resulted that there was a tendency of heterogeneous variance.Using weighted partial spline model with GMA method give better result than original partial spline model. This finding indicates the model of weighted partial spline using GMA method is better than original partial spline model in explaining the heterogeneity of variance.
关键词：Weighted Partial Spline; Generalized Moving Average; Original Spline
