A Stock Market Forecasting Model through Integration of Genetic Algorithm and Fuzzy Neural Network
碩士 === 義守大學 === 管理科學研究所 === 85 === Stock market is a very complicated environment which may vary in every minute. In addition, a large number of factors which can affect the stock market. Thus, it is very difficult to estimate the stock market tendency accurately via a single person''s k...
Main Author: | |
---|---|
Other Authors: | |
Format: | Others |
Language: | zh-TW |
Published: |
1997
|
Online Access: | http://ndltd.ncl.edu.tw/handle/09652428968038356555 |
id |
ndltd-TW-085ISU03457011 |
---|---|
record_format |
oai_dc |
spelling |
ndltd-TW-085ISU034570112015-10-13T12:15:16Z http://ndltd.ncl.edu.tw/handle/09652428968038356555 A Stock Market Forecasting Model through Integration of Genetic Algorithm and Fuzzy Neural Network 應用遺傳演算法與模糊神經網路於股票預測模式之研究 黃永成 碩士 義守大學 管理科學研究所 85 Stock market is a very complicated environment which may vary in every minute. In addition, a large number of factors which can affect the stock market. Thus, it is very difficult to estimate the stock market tendency accurately via a single person''s knowledge. Though many stock experts have dedicated in this area for many decades, so far, it is still an open question. Therefore, this research does not only dedicate to collect the qualitative factors but also the quantitative factors in order to overcome the above concern Then apply them to train the proposed fuzzy neural network (FNN) and artificial neural networks (ANN). Through such procedures, the trained network is a dynamic network with learning, inference, and forecasting capabilities. Regarding the qualitative factors, this research divides them int six dimensions, political, economical, financial, international, and message. For each dimension, all the factors are further grouped in order to form the fuzzy IF-THEN rules which are used to train the FNN. The proposed FNN learning algorithm consists of two components : (1) genetic algorithm and (2) error backpropagation (EBP) type learning algorithm. The reason to employ the genetic algorithm is to avoid reaching the local minimum and speed up the training. Finally, the affective levels from six dimensions, or FNNs, are integrated with the technical index, or quantitative factors, in order to train the feedforward neural network with EBP learning algorithm. The forecasting results showed that the proposed intelligent forecasting system with 3.86% MSE value outperforms the single ANN. 郭人介 鄭駿豪 1997 學位論文 ; thesis 155 zh-TW |
collection |
NDLTD |
language |
zh-TW |
format |
Others
|
sources |
NDLTD |
description |
碩士 === 義守大學 === 管理科學研究所 === 85 === Stock market is a very complicated environment which may vary in every minute. In addition, a large number of factors which can affect the stock market. Thus, it is very difficult to estimate the stock market tendency accurately via a single person''s knowledge. Though many stock experts have dedicated in this area for many decades, so far, it is still an open question. Therefore, this research does not only dedicate to collect the qualitative factors but also the quantitative factors in order to overcome the above concern Then apply them to train the proposed fuzzy neural network (FNN) and artificial neural networks (ANN). Through such procedures, the trained network is a dynamic network with learning, inference, and forecasting capabilities.
Regarding the qualitative factors, this research divides them int six dimensions, political, economical, financial, international, and message. For each dimension, all the factors are further grouped in order to form the fuzzy IF-THEN rules which are used to train the FNN. The proposed FNN learning algorithm consists of two components : (1) genetic algorithm and (2) error backpropagation (EBP) type learning algorithm. The reason to employ the genetic algorithm is to avoid reaching the local minimum and speed up the training.
Finally, the affective levels from six dimensions, or FNNs, are integrated with the technical index, or quantitative factors, in order to train the feedforward neural network with EBP learning algorithm. The forecasting results showed that the proposed intelligent forecasting system with 3.86% MSE value outperforms the single ANN.
|
author2 |
郭人介 |
author_facet |
郭人介 黃永成 |
author |
黃永成 |
spellingShingle |
黃永成 A Stock Market Forecasting Model through Integration of Genetic Algorithm and Fuzzy Neural Network |
author_sort |
黃永成 |
title |
A Stock Market Forecasting Model through Integration of Genetic Algorithm and Fuzzy Neural Network |
title_short |
A Stock Market Forecasting Model through Integration of Genetic Algorithm and Fuzzy Neural Network |
title_full |
A Stock Market Forecasting Model through Integration of Genetic Algorithm and Fuzzy Neural Network |
title_fullStr |
A Stock Market Forecasting Model through Integration of Genetic Algorithm and Fuzzy Neural Network |
title_full_unstemmed |
A Stock Market Forecasting Model through Integration of Genetic Algorithm and Fuzzy Neural Network |
title_sort |
stock market forecasting model through integration of genetic algorithm and fuzzy neural network |
publishDate |
1997 |
url |
http://ndltd.ncl.edu.tw/handle/09652428968038356555 |
work_keys_str_mv |
AT huángyǒngchéng astockmarketforecastingmodelthroughintegrationofgeneticalgorithmandfuzzyneuralnetwork AT huángyǒngchéng yīngyòngyíchuányǎnsuànfǎyǔmóhúshénjīngwǎnglùyúgǔpiàoyùcèmóshìzhīyánjiū AT huángyǒngchéng stockmarketforecastingmodelthroughintegrationofgeneticalgorithmandfuzzyneuralnetwork |
_version_ |
1716856279526277120 |