Smoothing based on stretched interpolated moving average approach

ABSTRACT: In this thesis, some smoothing techniques in multivariate and functional data analysis such as, kernel smoothing, local linear regression (LLR), spline smoothing and smoothing together with principal components analysis through conditional expectation (PACE) methods are considered. Their details are studied and a new smoothing method benefiting from moving average concept and applicable under certain conditions is proposed. Due to the steps involved in its logic, the proposed method is named Strecthed Interpolated Moving Average (SIMA). Its application to different data sets produced better results in terms of involved error, compared with LLR and similar results when compared with PACE. Keywords: Karhunen–Loève Expansion, Stretched Interpolated Moving Average, Principal Component Scores, Lag Interval, Weight Function. …………………………………………………………………………………………………………………………………………………………………………………………………………