🌌 Team Arjuna’s Journey at NASA Space Apps Challenge 2024: Solar Flare Energy Prediction | RHESSI Mission 🚀

The NASA Space Apps Challenge 2024 provided us with a platform to push the boundaries of innovation in space research, and Team Arjuna embraced this opportunity with enthusiasm. In this blog, we share our journey, challenges, and achievements as we built a predictive model for solar flare energy ranges using machine learning techniques, contributing to the RHESSI Mission.

Project Overview: Predicting Solar Flare Energy with Machine Learning

Our challenge was to create a predictive model to analyze and forecast the energy ranges of solar flares. Solar flares are powerful bursts of radiation from the sun, and predicting their intensity and occurrence is crucial for mitigating their potential impact on Earth’s technological systems. Our approach combined data analysis with advanced machine learning techniques to deliver accurate and reliable predictions.

Key Phases of the Project

1. 🔍 Analysis Phase

• Understanding Solar Flares: We began by studying the behavior of solar flares, focusing on their energy distribution, duration, and frequency of occurrence.
• Feature Extraction: By identifying patterns in historical data, we carefully selected features that were most relevant to building robust predictive models.

2. 💡 Building Predictive Models

We experimented with multiple machine learning algorithms to identify the best models for predicting solar flare energy ranges:

• Linear Regression
• Logistic Regression
• Decision Tree
• Random Forest
• K-Nearest Neighbors (KNN)
• Gradient Boosting
• Neural Network

📊 Top-Performing Models and Their Evaluation

After evaluating each model, here are the top performers based on accuracy:

• Gradient Boosting Classifier: Achieved an accuracy of 87%
• Random Forest Classifier: Achieved an accuracy of 86%
• Decision Tree Classifier: Achieved an accuracy of 82%

To ensure our models provided meaningful insights, we rigorously assessed them using metrics like Accuracy, Precision, Recall, and F1-score.

Key Insights from Our Analysis

• Gradient Boosting and Random Forest emerged as the most reliable models, demonstrating the potential to predict solar flare energy ranges with high precision.
• Our findings represent a significant step forward in space weather research, offering a valuable tool for better understanding and predicting solar flare activity.

Innovative Web App Development

To further enhance our project’s impact, we developed a web app that integrates live data from NASA’s open datasets into our models. This app continuously updates the dataset, allowing our predictions to become more accurate over time. This dynamic approach ensures that our predictive capabilities evolve alongside new data, making the system adaptable to future developments.

Our Experience at the NASA Space Apps Challenge 2024

A Learning Journey at Nirma University, Ahmedabad

• Although we didn’t secure the top spot, being among the top 18 finalists and achieving 5th place was a testament to our team’s dedication and innovative spirit.
• Presenting our project to aerospace experts, researchers, and scientists was an invaluable experience that challenged us to refine our ideas and think critically about our approach.

Recognition and Feedback

• Our project was praised for its potential to contribute to ongoing space weather research and its innovative use of live data integration for real-time predictions.
• The feedback from industry experts provided us with new perspectives on improving our models and scaling our web app for broader use.

Acknowledgments

Our journey wouldn’t have been possible without the incredible support and guidance of the event mentors, judges, and the organizing team at Nirma University. We are especially grateful to Anil Bhardwaj, Director of the Physical Research Laboratory, whose inspiring words at the event’s inauguration fueled our motivation.

• Mentors: Ankit D., Mohit Rohilla, Kush Patel, Gaurav Chopda, and others provided crucial insights that helped us refine our approach.
• Judges: Ranendu Ghosh(Founder at SatLeo Labs | Retired Scientist, ISRO), Dr. Parul Patel( Former Group Director at ISRO — Indian Space Research Organization), and the panel of judges offered invaluable feedback that motivated us to strive for excellence.

Conclusion: Reaching for the Stars

The NASA Space Apps Challenge 2024 was more than just a competition for us — it was a learning experience that shaped our approach to problem-solving and innovation. We are excited to take our project further and explore new opportunities to collaborate and contribute to space weather research.

Follow Our Work

Check out our GitHub project for the Solar Flare Energy Prediction Model here. Stay tuned as we continue to refine our models and enhance our web app for even better predictions.

Thank you for joining us on this journey, and we look forward to pushing the boundaries of innovation in space research. Let’s keep reaching for the stars! 🌠

#NASA #SpaceAppsChallenge #SolarFlarePrediction #MachineLearning #DataScience #AI #SpaceWeather #RHESSIMission #TeamArjuna #Innovation #Research #STEM