IMPLEMENTING GENETIC ALGORITHM AT SEARCHING FOR PARAMETER OF EMPI WITH 17 INDICATOR

• Main Author: Gunawan (NIM : 13502072), Felix
• Format: Final Project
• Language: Indonesia
• Online Access: https://digilib.itb.ac.id/gdl/view/7903
• Institution: Institut Teknologi Bandung

Abstract: <br /> <br /> <br /> <br /> <br /> <br /> In this final project, has been developed an application. This application is searching for parameter of EMPI with 17 indicators. EMPI or Exchange Market Pressure Index is a formula for detecting any chance that a crisis will occur in the future. In present, the only EMPI formula which is in used is static formula. The newest development of this formula is using neural network for searching EMPI formula. In this final project, genetic algorithm will be used for searching a parameter for EMPI with 17 indicators. In this genetic algorithm, there are three important techniques. There are mutation technique, crossover technique, and selection technique. The mutation technique will be used is point mutation. The crossover technique will be used is two-point crossover. The selection technique will be used is roulette wheel selection with fix number of population. The most important functionality which given by this application is the function to help testing the performance, especially if it compares with neural network. This functionality give the end user the number of EMPI which the result of the best parameter EMPI according this application. Other functionality is to give the mount of fitness for each best chromosome form each generation. Before building this application, domain analysis according EMPI problem has been done. Next, analysis, plan, implementation, application testing has been done. Unified Modelling Language is used for building this application. This application is application base by .Net Framework build by using Visual Basic .Net 2003. According to the test, application has been fulfil compatibility specification with plan and can work according to the scenario that has been planned. This application, other than can be used for looking parameter of EMPI, it still can be developed so it will increase applications performance. The weakness of this application is the performance other than not better than using neural network is the time is need to accomplish the process. <br /> <br />