PARAMETER ESTIMATION AR AND ARMA TIME SERIES MODEL BY GAUSS-NEWTON METHOD AND GENETIC ALGORITHM. CASE STUDY: WATER DISCHARGE SAGULING RESERVOIR - WEST JAVA

• Main Author: FERDIANSYAH NUGRAHA, RIDA
• Format: Final Project
• Language: Indonesia
• Online Access: https://digilib.itb.ac.id/gdl/view/19308
• Institution: Institut Teknologi Bandung

The forecasting of hydrological variable, such as water discharge, has important role in planning and managing the water resources. A reservoir of water discharges that are recorded based on time will form a time series observations. This nal project conducts two time series modeling methods to the water discharge data of Saguling, West Java. Those methods are the Box-Jenkins method and numerical methods method. The numerical methods consist of the genetic algorithm and the Gauss-Newton. Both of methods showed that the AR(1) and ARMA(1,1) are the best models. Furthermore, by using AR(1) and ARMA(1,1) stationary models, the next few months water discharge can be forecasted.