| dc.description.abstract | Renewables are taking up the role of increasing importance in the energy industry, especially with the recent wave of the drop in prices of oil and gas. Therefore, oil and gas companies are gradually positioning themselves for the proclaimed energy transition. This raises the question of when it is a good time for the oil and gas companies to invest in renewable energy. This research helps to address this query by investigating the past behaviour of oil majors’ in terms of investment into renewable energy. Furthermore, the dependency of investment in renewable energy and oil and gas prices will be formulated and quantified, followed by a forecast of investment in renewable energy based on the forecast of oil and gas prices. Then, the impact of future investment in renewable energy and oil and gas prices on relevant investment decisions will be investigated. The objective is to build a renewable energy decision support system that predicts the investment decision by using crude oil prices and renewable energy stocks.
In this research, we used the Facebook prophet model that first trained on the historical data, then forecasts the crude oil and renewable energy stocks prices. The tree-based pipeline optimization method is used to select the classification model and its hyperparameters. The trend in historical data indicated that the ensemble of naive Bayes and extreme gradient boosting approach with randomized singular value decomposition principle component analysis was able to achieve 0:92 ROC. The able to forecast the short-term and long-term investments with an average classification F measure of 0:88. The Prophet forecasted model achieve mean square for crude oil, Frist solar, Brook field renewable, and Solar edge technologies 13:69; 27:74; 6:67; 64:04.
The proposed method can infer the Prophet and TPOT based method better
understand the crude oil and renewable energy sectors. The proposed approach can over/underestimates high or low peaks as forecasted with mean square error. Therefore, it is recommended to use additional features, i.e. momentum and trend analysis, news sentiments, and market indicators with lagging information. The proposed method can be used as an indicator for oil and renewable energy. | |