Chaos for Rare Earth Elements Price forecasting-An Analysis of Artificial Neural Networks

碩士 === 中原大學 === 企業管理研究所 === 100 === This study aims to predict the price of the Rare Earth Elements (REE), which are key components of green energy technologies and other high technology applications. The use of REEs in modern technology has increased dramatically over the past years. Based on the g...

Full description

Bibliographic Details
Main Authors: Tushigmaa Batsukh, 圖史格
Other Authors: Jo Hui Chen
Format: Others
Language:en_US
Published: 2012
Online Access:http://ndltd.ncl.edu.tw/handle/56124205566294384217
Description
Summary:碩士 === 中原大學 === 企業管理研究所 === 100 === This study aims to predict the price of the Rare Earth Elements (REE), which are key components of green energy technologies and other high technology applications. The use of REEs in modern technology has increased dramatically over the past years. Based on the growing global demand for REEs and the limited supply, there is growing concern that the world may soon face a shortage of the materials. As a result, prices have risen significantly. A rise in prices in REEs market creates challenging for mining and manufacturing companies in pricing of products and services because of high cost raw material. To reduce REE explosion risk and make hedging, future investment and evaluation decision will depend on accurately forecasting future price trend. This study applied three different approaches, Brock-Dechert Scheinkman test, Rescaled range analysis and Correlation Dimension Analysis for detection of chaotic phenomena. Next this study utilizes the artificial neural networks, including back propagation network (BPN) and Time delay recurrent neural network (TDRNN) to make a prediction for the price of REEs associated with the inputs such as The Broad Index, Baltic Dry Index, Commodity Research Bureau Futures Price Index, PHLX Semiconductor Sector index, NASDAQ Computer Index and The London Interbank Offered Rate. The result will explore the price behavior behind the REEs and to provide investors by the valuable information, this study compares which network forecast is more accurate. The simulation resulted that the chaos effect is exists in REE prices and suggest that employing TDRNN in price data of REE is more effective than BPN and the best performance is attained by the TDRNN. Therefore, REEs associated inputs, The Broad Index, Baltic Dry Index, Commodity Research Bureau Futures Price Index, PHLX Semiconductor Sector index, NASDAQ Computer Index and The London Interbank Offered Rate would be good indicator for forecasting REEs.