Nonlinear Modeling of IHSG with Artificial Intelligence

Abstract

Artificial Intelligence is the simulation of human intelligence processes by computer systems which can be used to model stock prices. Learning algorithms of artificial neural network used to train the network so far the weight of connection inter units can be suitable with error which have determined. The back propagation method is designed as operation of feed-forward network with multiple layers in order that the result of the weights is nonlinear. Nonlinear weights make a nonlinear model in artificial neural network. Time series data of Composite Stock Prices Index (IHSG) is trained using back propagation method in artificial neural network until error which is obtained in weights of the network become very small. The weights is used to model IHSG. Performance rate of time series data model of IHSG which started on January 2016 until December 2017 is measured using Mean Absolute Percentage Error (MAPE). Based on MAPE value of 1.74528596% indicates that the model obtained is very good used to forecast IHSG in the future.