Open AccessPrediction of Jakarta Composite Index Using Neural Network Model and Genetic Optimization
Author(s) -
Rukun Santoso,
Budi Warsito,
Huda Yasin
Publication year - 2020
Publication title -
journal of physics. conference series
Language(s) - English
Resource type - Journals
SCImago Journal Rank - 0.21
H-Index - 85
eISSN - 1742-6596
pISSN - 1742-6588
DOI - 10.1088/1742-6596/1655/1/012096
Subject(s) - conjugate gradient method , gradient descent , artificial neural network , computer science , genetic algorithm , heuristic , gradient method , sample (material) , nonlinear conjugate gradient method , data mining , artificial intelligence , algorithm , machine learning , chemistry , chromatography
Researches related to prediction of stock price data have been developing rapidly. Likewise, the modeling techniques used for predictive purposes are also increasing along with advances in the field of computing. This study applied neural network model in predicting the Jakarta Composite Index data as a case of time series. The optimization method used was genetic algorithm. This method is included in one of the heuristic techniques. Unlike standard optimization methods, genetic algorithms do not use gradients as a basis for search techniques. Parameters in the neural network model are obtained from the process of decoding chromosomes from generation to generation. In comparison, the two gradient-based optimization methods were also applied, i.e Conjugate Gradient and Gradient Descent. The results showed the superiority of genetic algorithms compared to other optimization methods in out-sample prediction whereas, the in-sample prediction of gradient-based optimization methods achieve better precision.