Neuro – Genetic model for the projection of crude oil price capable of handling of uncertainty / Haruna Chiroma
Some events occur sometimes without any warning, such as war, revolution, financial crises, terrorist attacks, political conflicts, false news, natural disasters, earthquakes, and extreme weather conditions. These types of events which we termed as uncertain events, when related to crude oil have si...
Saved in:
| Main Author: | |
|---|---|
| Format: | Thesis |
| Published: |
2015
|
| Subjects: | |
| Online Access: | http://studentsrepo.um.edu.my/6836/1/THESIS_PRELIMINARY_PAGES.pdf http://studentsrepo.um.edu.my/6836/2/PHD_THESIS.pdf http://studentsrepo.um.edu.my/6836/ |
| Tags: |
Add Tag
No Tags, Be the first to tag this record!
|
| Summary: | Some events occur sometimes without any warning, such as war, revolution, financial crises, terrorist attacks, political conflicts, false news, natural disasters, earthquakes, and extreme weather conditions. These types of events which we termed as uncertain events, when related to crude oil have significant effects on the price and will contribute to oil price volatility. Volatility in crude oil market has direct and indirect negative effects on the global economy and inflicts suffering on communities across the globe. The effects of crude oil volatility have no geographical boundary as there is no restriction to a specific country or region of the world. The purpose of the research is to propose a model that can predict the price of crude oil in the real world scenario. This study presents an alternative model based on Neural Network and Genetic Algorithm (Neuro-Genetic) for the projection of crude oil price while considering the impact of uncertainties. The difference between the crude oil price projected by the Neuro-Genetic model and the actual price was not statistically significant. The results obtained by the Neuro-Genetic model performs significantly better than the backpropagation neural network and support vector machine in both accuracy and CPU processing time. The model was able to learn patterns from volatile crude oil price datasets during the 1991 Gulf War, the 1997 Asian financial crisis, the 2002 Venezuelan unrest, the second Gulf War of 2003, and the 2007 global financial recession. The retraining applied in the modeling process possibly allow the Neuro-Genetic model to learn and capture new data patterns during the uncertain events. Thus, the model can effectively be applied as an alternative mechanism by policy makers in the formulation of policies related to energy demand and supply, bio-fuel, fuel subsidy, global food price subsidy, the stock market as well as national planning and budget. Intergovernmental organizations such as the Organization of Petroleum Exporting Countries (OPEC) can use our proposed model to serve as a guide for the formulation of policies related to international crude oil price. The model has the potential for realistic, practical application in the real world. |
|---|