PREDICTION MODEL INAR(1) FOR THE HEALTH STATUS

• Main Author: SIHOMBING (NIM : 10108107); Pembimbing : Khreshna I.A. Syuhada, M.Sc, Ph.D, FEBRINA
• Format: Final Project
• Language: Indonesia
• Online Access: https://digilib.itb.ac.id/gdl/view/15657
• Institution: Institut Teknologi Bandung

Integer-Valued Autoregressive (INAR) is a time series prediction model for counting data. In this Final Project, we employ a Poisson INAR model of order one to predict a child's health status at time t + 1 based on two components: the health status at time t and the unexpected health status change at time (t, t+1). The INAR(1) model has same characteristics as compared to the classical AR(1) model in the context of (i) stationary property, and (ii) distributional property that is determined by its error distribution. We have used the methods of least square and maximum likelihood in order to estimate model parameter. Simulation study is carried out to illustrate and calculate a prediction limit of the model, in which its prediction <br /> <br /> <br /> <br /> <br /> accuracy is assessed by calculating conditional mean square error of prediction and coverage probability.