Predicting Cometary Pathway using Linear Regression

Making space science accessible—predicting comet paths using simple, open-source machine learning.

Comment

Inspiration

The inspiration for this project came from a desire to make space science more accessible to students and researchers without advanced computational resources or astrophysics backgrounds. While exploring orbital prediction, I realized that most existing tools are complex and domain-specific. I wanted to simplify this process using machine learning, so anyone with basic coding skills could model comet trajectories.

What it does

The project uses linear regression to predict the future positions of 1P/Halley’s Comet based on its historical trajectory data. It demonstrates that even simple, interpretable ML models can be applied to celestial mechanics and highlights the importance of standardizing evaluation methods for ML-based orbit prediction tools.

How I built it

  1. I developed the project as part of an academic assignment under the mentorship of my professor. I:

  2. Collected historical orbital data for 1P/Halley

  3. Preprocessed and structured the data

  4. Trained a linear regression model using Python

  5. Visualized the predicted vs. actual positions

Proposed a framework to encourage standardized evaluation and interoperability for ML in orbit prediction

Challenges I ran into

  1. Handling non-linear and chaotic nature of comet trajectories with a linear model

  2. Ensuring the prediction remained meaningful despite model simplicity

  3. Accessing clean, time-series ephemeris data for training

  4. Proving the relevance of a lightweight model in a field dominated by complex simulations

Accomplishments that I'm proud of

  1. Created a working ML model for comet path prediction using only open-source tools

  2. Published the work on TechRxiv, opening it up for peer feedback and community engagement

  3. Proposed a need for interoperability and benchmarking standards in the field

  4. Made orbital modeling more approachable for learners and low-resource environments

What I learned

  1. The power of simplicity and accessibility in scientific innovation

  2. Fundamentals of astronomical data handling and time-series prediction

  3. Importance of standardization in ML research, especially across scientific disciplines

  4. That impactful innovation doesn't always require complex technology—it often starts with clarity and purpose

What's next for Predicting Cometary Pathway using Linear Regression

  1. Extend the model to include non-linear regressors or neural networks

  2. Build a web-based visualization tool for comet predictions

  3. Collaborate with other researchers to develop a benchmark dataset and evaluation framework

  4. Adapt the model to predict paths of other short-period comets and asteroids

Built With

  • google-colab
  • linear-regression
  • python
  • sickit-learn
Share this project:

Updates