Exo Vision AI

• 

  The Exo Vision AI

Inspiration

1. Project Exo Vision AI (or whichever name you chose)2. Short TaglineSeparating real worlds from stellar noise in NASA space telescope data using AI.3. About the Project (Main Story / Description Field)🚀 Inspiration Finding new planets outside our solar system (exoplanets) requires sifting through massive amounts of space telescope observation data. Distinguishing true planetary transits from background noise or instrumental errors is a core challenge faced by astrophysicists every day. We created Exo Vision AI to automate this pipeline using modern machine learning, proving that high school students can contribute to real space data analysis using NASA's public archives.🔍 What It Exo Vision AI is an end-to-end data science and machine learning pipeline built on observational data sourced directly from the NASA Exoplanet Archive. It automatically pre-processes light-curve observational metrics, handles missing values, addresses target class imbalance, and classifies candidate signals into either true exoplanets or false positives with high accuracy and F1-score.🛠️ How We Built It Data Processing & EDA: Handled missing values using median imputation, scaled features, and analyzed correlations across transit depth, orbital period, and planetary radius using pandas and seaborn. Machine Learning Pipeline: Tested and compared multiple classification algorithms—including Random Forest and Gradient Boosting (XG Boost)—evaluated via K-Fold cross-validation. Evaluation & Visuals: Generated confusion matrices, ROC-AUC curves, and feature importance plots to analyze how the model makes predictions.📊 Model Performance Metric Result Accuracy[Insert e.g., 94.2%]Precision[Insert e.g., 93.5%]Recall[Insert e.g., 92.8%]F1-Score[Insert e.g., 93.1%]🧠 Key Insights & Learnings Top Predictive Features: Observational metrics like Transit Depth (ppm) and Planetary Radius had the highest feature importance scores in determining true candidates. Non-Technical Explanation: Just like filtering out radio static to hear a clear song, our model learns the unique "shadow signature" a planet casts when passing in front of its star, ignoring random fluctuations in starlight.🔮 What's Next for Exo Vision AI We plan to scale this model to process raw light-curve time-series data using 1D Convolutional Neural Networks (CNNs) and expand its pipeline to handle upcoming data streams from the James Webb Space Telescope (JWST).4. Built With python Jupyter Notebook Scikit-Learn Pandas NumPy Matplotlib Seaborn GitHub ## What it does

How we built it

Challenges we ran into

Accomplishments that we're proud of

What we learned

What's next for Exo Vision AI

Built With

• ai
• artificial-intelligence
• astronomy
• astrophysics
• classification
• data-cleaning
• data-science
• data-visualization
• eda
• exoplanet
• github
• jupyter
• kepler
• machine-learning
• nasa
• numpy
• pandas
• python
• random-forest
• scikit-learn
• space
• space-tech
• stem
• xgboost