PENGGUNAAN KOMPONEN CHAOS DAN ARTIFICIAL NEURAL NETWORK UNTUK PERAMALAN HARGA SAHAM

• Main Author: Subron Muhyiddin, Dandy
• Format: Theses
• Language: Indonesia
• Online Access: https://digilib.itb.ac.id/gdl/view/3188
• Institution: Institut Teknologi Bandung

<b>ABSTRACT:</b><br> <br /> <br /> USING CHAOS COMPONENT AND ARTIFICIAL NEURAL NETWORK <br /> FOR STOCK PRICES FORECASTING <br /> <br /> by : <br /> Dandy Sobron Muhyiddin NIM : 230 02 029 <br /> <br /> <br /> This study is an empirical investigation of the some particular stock prices behaviour at Jakarta Stock Exchange, describes through the frame work of non linear dynamic or Chaos Theory and perform forecasting with artificial neural network (ANN). Chaos theory analyzes process under assumption that a part of process is random and a part is deterministic. With this assumption forecasting technique is possible which is contrary to the efficient market hypothesis.</p> <br /> <br /> The study examined the presence of the chaotic system, characterized by positive Lyapunov exponent and fractal dimension. Through fractal dimension can be determined another chaos component, embedding dimension, considered assist information about how many time lags are involved in the ANN forecasting. ANN is claimed able to model both deterministic and random process, so it is considered ideal for non linear dynamic system forecasting.</p> <br /> <br /> The result showed that chaotic deterministic characteristic are found for all data observation. Generally, performing ANN forecasting showed good performance, and using chaotic component embedding dimension indicated assist the forecasting result. <br />