/StarryNight

A Python notebook based on the Hertzsprung-Russell Diagram, which employs machine learning methodologies to unveil inherent patterns within stellar data, aiming to assist scientists and astronomers.

Primary LanguageJupyter NotebookMIT LicenseMIT

StarryNight

Team Members

Overview

The StarryNight project is a Python notebook that leverages the Hertzsprung-Russell Diagram (HR-Diagram) and employs machine learning methodologies to uncover inherent patterns within stellar data. The primary goal of this project is to establish a robust framework for categorizing stars based on their discernible features, ultimately aiding scientists and astronomers in their research endeavors.

Features

  • Utilizes Python programming language.
  • Incorporates machine learning techniques for stellar data analysis.
  • Focuses on the Hertzsprung-Russell Diagram for star classification.
  • Provides a systematic approach to categorize stars based on observable features.

Objective

The overarching objective of the StarryNight project is to construct a star classification system using machine learning techniques acquired during the course of the Machine Learning curriculum. By doing so, the project aims to demonstrate that stars adhere to a discernible pattern, specifically manifested in the HR-Diagram. This diagram serves as the foundation for classifying stars by plotting their features, offering valuable insights into their characteristics.

Project Structure

Domain Description and Project Objectives

This project delves into the intricate realm of stellar phenomena with the primary goal of constructing a robust classification system for stars based on their positions on the Hertzsprung-Russell Diagram. The overarching objectives include gaining profound insights into stellar characteristics and leveraging machine learning to enhance our understanding of celestial bodies.

Design Choices and Dataset Creation

  1. Domain Exploration: Deliberate design choices have been made to meticulously curate a dataset that captures the essence of stellar properties. Assumptions and hypotheses underpin the data collection strategy, ensuring a comprehensive representation of celestial phenomena.

  2. Data Set Description and Exploratory Analysis: A detailed overview of the curated dataset is provided, accompanied by an in-depth exploratory analysis. This step aims to uncover patterns, trends, and anomalies within the stellar data, laying the groundwork for subsequent machine learning endeavors.

Machine Learning Model Selection and Rationale

  1. Machine Learning Model Choices: Two distinct machine learning models have been strategically chosen to address the classification challenge. The rationale behind each model selection is elucidated, considering their strengths and suitability for discerning patterns within the Hertzsprung-Russell Diagram.

Experiments and Performance Evaluation

  1. Experimental Framework: Rigorous experiments have been conducted, incorporating at least one validation methodology to assess the efficacy of the proposed classification system.

  2. Performance Metrics: The performance of the implemented models is meticulously evaluated, with a focus on measuring key metrics that gauge the accuracy and reliability of the classification results.

Results Analysis

  1. Insights and Reflections: A comprehensive analysis of the obtained results is presented, offering valuable insights into the effectiveness of the classification system. Patterns and correlations discovered within the stellar data are thoroughly examined, contributing to the broader understanding of celestial bodies.

  2. Conclusions: This section encapsulates the final reflections and implications drawn from the project. It discusses the significance of the findings, potential avenues for further research, and the broader impact of the developed classification system on the field of astrophysics.

License

This project is licensed under the MIT License.

Made With ❤️ in Milan, Bicocca