We built a sophisticated long short-term memory neural network (LSTM) machine learning model specifically designed to predict the number of barrels of crude oil received each month for the next two years by Exxon Mobil Corporation. This undertaking required a comprehensive understanding of both the intricacies of machine learning models and the fluctuations in the oil market. To facilitate our analysis, we utilized the U.S. Energy Information Administration's open data API, which provided us with a rich dataset containing historical data on Exxon Mobil Corporation's crude oil imports, dating back to January 2009. This dataset served as the foundation for our predictive modeling efforts, allowing us to glean insights into past import behaviors and trends. Below is a graph depicting the predicted crude oil imports for Exxon Mobil Corporation for 2025 through 2026.
An LSTM neural network is a specialized type of recurrent neural network (RNN) that excels in processing sequential data. This capability makes it particularly well-suited for tasks involving time-series forecasting, where understanding long-term dependencies is crucial for making accurate predictions. In our project, we meticulously trained the LSTM model using a comprehensive dataset that included historical records of crude oil imports, specifically focusing on the number of barrels imported each month by Exxon Mobil Corporation. By leveraging this extensive historical data, the LSTM model was able to learn intricate patterns and trends that characterize the fluctuations in oil imports over time. Below we show a graphical depiction of the performance of the LSTM neural network.
Through this training process, we aimed to develop a robust predictive framework that could not only forecast future import quantities for Exxon Mobil but also identify any emerging trends, such as seasonal variations or quarterly cycles that might influence these imports. For instance, we sought to determine whether there are consistent patterns in the data that indicate increases or decreases in imports depending on the time of year, which could be influenced by factors such as market demand, geopolitical events, or changes in domestic production levels. By analyzing these trends, our model aims to provide valuable insights into the operational strategies of Exxon Mobil Corporation and the broader dynamics of the crude oil market.
Ultimately, our goal is to create a reliable forecasting tool that can assist stakeholders in making informed decisions based on predicted import volumes, thereby enhancing their understanding of market dynamics and facilitating strategic planning. The insights generated from our LSTM model could prove invaluable for Exxon Mobil Corporation as it navigates the complexities of the global oil market over the coming years.
Comentários