Skip to content

Release Notes - The Oval Table

The Oval Table - Oil Price Prediction Models

Version: v1.0

Cycle 1 (08/24/2023 - 10/08/2023) Summary:

Introduction to ML:

  • First venture into the realm of Machine Learning.

  • Laid the groundwork with Regression models to predict oil prices.

  • Models Introduced:

    • XGBoost: Utilized a gradient boosting framework with decision trees.

    • Random Forest: An ensemble learning method utilizing multiple decision trees.

    • Polynomial Regression: A regression algorithm modeling the relationship between input and output as an nth degree polynomial.

    • Baseline Model: Introduced a Naive Forecast as a basic comparison tool.

Challenges & Learnings:
  • Overfitting: Identified that models were prone to overfitting on the training data.
Next Steps:
  • Implement model refinement to address overfitting issues.
  • Integrate time series forecasting to predict future oil prices.
  • Web interface integration.

Cycle 2 (10/09/2023 - Present) Updates:

Shift in Focus:

  • Transitioned the core focus towards Time Series models, as per the wants and requirements of our 'client', (a professor at CSU).

  • A priority has been set on enhancing oil price prediction accuracy using Time Series methods.

Time Series Models:

  • Delving into Time Series models to cater to the specific needs of oil price forecasting.

  • Research and development are in progress to refine these models and improve their predictive capabilities.

Model Refinement Plans:

  • While Time Series remains the primary focus, there's an underlying aim to address the overfitting issues from Cycle 1.

  • Depending on the progress and time constraints, there might be further refinements made to the regression models from the previous cycle.

Data Cleaning & Preprocessing:

  • Recognized the need for more meticulous data cleaning.

  • Plans are set to further refine preprocessing steps, if the semester's timeline allows.