RISK PREDICTION USING VARIANCE

• Main Author: Nessa, Widya
• Format: Theses
• Language: Indonesia
• Online Access: https://digilib.itb.ac.id/gdl/view/47892
• Institution: Institut Teknologi Bandung

Variance as a risk prediction represents the distribution of risk data relative to the mean. Calculating variance of risk data using risk random variable needs to be considered to obtain the value of variance in all types of risk data. Variance of random variable is obtained by using the probability function of the risk data distribution. If the risk data is bivariate data, the variance which can be calculated is conditional variance. Conditional variance becomes dicult to be counted when certain standard joint probability functions, such as the bivariate normal and bivariate Student's t, do not match the bivariate risk data. For bivariate models with marginal distributions from dierent distribution families, the joint probability function can be determined using the Bivariate Clayton Copula. The bivariate variance values of various marginal distribution choices can be determined using joint probability function of the Bivariate Clayton Copula. The analysis is based on normal bivariate simulation data and another bivariate model which has Weibull and Exponential marginal distributions. In addition, an analysis was also conducted on Angka Partisipasi Sekolah (APS) data from 34 provinces in Indonesia in the period 2011 to 2019. From these two analyzes the prediction of on (1 ???? ) condence were determined. Based on the estimated coverage probability and back????testing test methods, risk prediction using variance provides fairly accurate prediction results.