Developing a hybrid hidden MARKOV model using fusion of ARMA model and artificial neural network for crude oil price forecasting
Crude oil price forecasting is an important component of sustainable development of many countries as crude oil is an unavoidable product that exist on earth. Crude oil price forecasting plays a very vital role in economic development of many countries in the world today. Any fluctuation in cr...
Saved in:
| Main Author: | |
|---|---|
| Format: | Thesis |
| Language: | English English English |
| Published: |
2020
|
| Subjects: | |
| Online Access: | http://eprints.uthm.edu.my/924/1/24p%20NUHU%20ISAH.pdf http://eprints.uthm.edu.my/924/2/NUHU%20ISAH%20COPYRIGHT%20DECLARATION.pdf http://eprints.uthm.edu.my/924/3/NUHU%20ISAH%20WATERMARK.pdf |
| Tags: |
Add Tag
No Tags, Be the first to tag this record!
|
| Summary: | Crude oil price forecasting is an important component of sustainable development of
many countries as crude oil is an unavoidable product that exist on earth. Crude oil
price forecasting plays a very vital role in economic development of many countries
in the world today. Any fluctuation in crude oil price tremendously affects many
economies in terms of budget and expenditure. In view of this, it is of great concern
by economists and financial analysts to forecast such a vital commodity. However,
Hidden Markov Model, ARMA Model and Artificial Neural Network has many
drawbacks in forecasting such as linear limitations of ARMA model which is in
contrast to the financial time series which are often nonlinear, ANN is very weak in
terms of out-sample forecast and it has very tedious process of implementation, HMM
is very weak in an in-sample forecast and has issue of a large number of unstructured
parameters. In view of this drawbacks of these three models (ANN, ARMA and
HMM), we developed an efficient Hybrid Hidden Markov Model using fusion of
ARMA Model and Artificial Neural Network for crude oil price forecasting,
MATLAB was employed to develop the four models (Hybrid HMM, HMM, ARMA
and ANN). The models were evaluated using three different evaluation techniques
which are Mean Absolute Percentage Error (MAPE), Absolute Error (AE) and Root
Mean Square Error (RMSE). The findings showed that Hybrid Hidden Markov Model
was found to provide more accurate crude oil price forecast than the other three
models in which. The results of this study indicate that Hybrid Hidden Markov Model
using fusion of ARMA and ANN is a potentially promising model for crude oil price
forecasting. |
|---|