FAQ
-
What is the main objective of this project?
Our primary goal is to leverage machine learning to predict oil prices with high precision. This project stems from our Software Engineering course (CPSC 4175), where we were paired with "clients" from within the academic institution. Our assigned 'client' stems from the economics department and is interested in contrasting the predictive capability of machine learning models against traditional econometric forecasting methods. Through this project, we aim to explore the synergies between technology and economics to achieve more accurate and insightful predictions.
-
Why did you start with Regression models in Cycle 1?
As newcomers to machine learning, we wanted to initiate our journey with a foundational understanding. Regression models, such as XGBoost, Random Forest, and Polynomial Regression, provided a suitable starting point. These models served as our initial attempt to understand machine learning concepts, oil price trends and correlations.
-
I noticed overfitting issues in Cycle 1. What are your plans to address this?
Yes, we identified overfitting during Cycle 1. In Cycle 2, while our primary focus is on Time Series models, we also aim to address and mitigate the overfitting issues, time permitting. We're actively researching best practices and fine-tuning techniques to improve model generalization. Remember, this is a learning experience for us too, so we're constantly evolving and adapting to new challenges.
-
Why the shift to Time Series models in Cycle 2?
Our 'client' expressed an interest in predicting future oil price trends, which is best suited to Time Series models. We're currently researching and developing Time Series models to improve their predictive capabilities.
-
Will there be further cycles after Cycle 2?
We're currently focused on delivering the best results in Cycle 2, especially with our semester ending in December. Future cycles or iterations will depend on various factors, including project outcomes, feedback, time and evolving objectives.
-
How often is the data updated?
The dataset spans from January 1986 to June 2023 for macroeconomic indicators and until July 2023 for oil prices. However, please note that this data isn't regularly updated, as it primarily serves the educational objectives of our school project.
-
Can I contribute or provide feedback?
Absolutely! Feedback is invaluable. We welcome contributions via pull requests on our [GitHub repository](). Additionally, for feedback, queries, or issues, you can use the "Issues" section on GitHub.