KAJIAN PEMODELAN SPLINE UNTUK DATA LONGITUDINAL SEBAGAI PERKEMBANGAN DARI REGRESI NONPARAMETRIK

Abstract

Regression analysis can be approached by using parametric, semi-parametric
and nonparametric regression approaches. One of nonparametric regression
approach that great developed was Spline truncated, including for modeling
longitudinal data. Longitudinal data is data that consisting of several subjects
which is each subject is observed repeatedly based on a certain time. The
advantages of longitudinal data has provided more complexcityof  information
than cross section and time series data. The spline approach was a segmented
polynomial regression approach. Spline provides high flexibility due to the use
of knot points. To determine the optimal knot points using Generalized Cross
Validation (GCV). The principle of determining the optimum point of knot of
longitudinal data using spline truncated is basically the same as with Spline
method  for cross section data, that is determination of knot point based on each
subject. However, the estimation is done simultaneously so that each subject has
its own model.
Keywords: Spline Truncated, GCV, Knot points.