🪐 E X O H A B I T A I

"Because the next Earth might already be out there — waiting to be found."

🌌 A full-stack machine learning system that predicts whether a distant exoplanet could harbor life —
powered by real astrophysical data, engineered features, and production-ready ML pipelines.

---

🌠 Mission Briefing

"There are more stars in the observable universe than grains of sand on Earth.
Among those stars orbit billions of planets. ExoHabitAI asks the question:
Which ones could be home?"

ExoHabitAI is not just a machine learning project — it is a scientific exploration tool.
It ingests raw exoplanet data, processes astrophysical signals, calculates Earth Similarity Indices, and deploys a trained ML model through a polished web dashboard to deliver habitability predictions in real-time.

This project bridges data science, astronomy, and software engineering — the way real-world AI systems are built.

---

🚀 Live Demo & Quick Access

Access Point	Link
🌐 Live App	b13-exohabitai-samridhigupta.onrender.com
🖥️ Local Dashboard	http://127.0.0.1:5000 after running Flask
🟢 Health Check	GET /
🔵 Model Info	GET /model-info
🚀 Predict	POST /predict
🏆 Top Rankings	GET /rank
📓 Notebooks	./notebooks/

---

🌌 Live Demo Experience

🚀 Watch ExoHabitAI in action — real-time prediction system

🎥 Demo Video:
👉 ▶️ Click to Watch

---

🖥️ Frontend Experience

A futuristic AI-powered interface designed for seamless space exploration and intelligent habitability analysis.

---

🚀 1. Initial Landing Experience

↑ First impression of ExoHabitAI — immersive space UI with strong visual identity and entry point into the system.

---

🌌 2. Homepage — System Overview

↑ Overview of the platform, showcasing AI-driven exoplanet research, data pipeline, and model performance.

---

🧠 3. Prediction Input & Top Ranked Planets

↑ Users input planetary parameters and instantly explore the top 20 most habitable exoplanets ranked by the model.

---

📊 4. Prediction Result Dashboard

↑ Detailed habitability report with probability score, parameter influence, and scientific insights.

---

🪐 5. About Page — Project Insight

↑ Explains the mission, methodology, and AI approach behind ExoHabitAI.

---

📈 6. Visualization Dashboard

↑ Advanced analytics dashboard with feature importance, distributions, model comparison, and data insights.

---

✨ Key Highlights

• 🌠 Futuristic space-themed UI/UX
• ⚡ Real-time AI habitability prediction
• 📊 Interactive data visualizations
• 🧪 Scientifically inspired feature engineering
• 🚀 Smooth navigation across multiple modules

---

🧬 Machine Learning Pipeline

From raw stellar data to intelligent predictions — a five-stage mission

RAW DATA  ──►  PREPROCESSING  ──►  FEATURE ENGINEERING  ──►  MODEL TRAINING  ──►  PREDICTION
   🪐                🧹                    🧮                       🤖                  ✅

Stage 1 — Data Ingestion & Cleaning

• Loaded the NASA Exoplanet Archive dataset with 30+ astrophysical features
• Removed duplicate records and irrelevant columns
• Standardized column naming and data types

Stage 2 — Missing Value Handling

• Applied median imputation for numerical orbital features
• Used domain-constrained bounds to cap outliers (e.g., stellar temperature ranges)
• Visualized missingness patterns using heatmaps

↑ Missing values heatmap — guiding imputation strategy

Stage 3 — Feature Engineering

• Engineered the Earth Similarity Index (ESI) from radius, density, escape velocity, and surface temperature
• Created orbital stability indicators
• Normalized all features using StandardScaler

Stage 4 — Model Training & Selection

Trained and evaluated five models:

Model	Accuracy	ROC-AUC
Logistic Regression	—	—
Decision Tree	—	—
Random Forest	—	—
XGBoost	—	—
Tuned Random Forest ✅	Best	Best

Replace — with your actual metrics from final_model_metrics.csv

Stage 5 — Evaluation

• Confusion matrices, ROC curves, and feature importance analyzed for every model
• Final model serialized as exohabit_model.pkl

---

📊 Visual Insights

The data tells a story — here's how we listened

Model Comparison

↑ Side-by-side accuracy and AUC comparison across all trained models

Feature Importance

↑ Top predictive features driving habitability classification

Confusion Matrices

Final Selected Model	Tuned XGBoost
Confusion Matrix — Best Model	Confusion Matrix — Tuned XGBoost

ROC Curves

Random Forest	Logistic Regression
Random Forest ROC Curve	Logistic Regression ROC Curve

---

⚙️ Backend & API

The engine room — where predictions are born

Architecture

Frontend (HTML/JS)
       │
       ▼
  Flask API  (app_deploy.py · port 5000)
       │
       ├──── GET  /              →  Health Check
       ├──── GET  /model-info    →  Model Metadata
       ├──── POST /predict       →  Habitability Prediction
       └──── GET  /rank          →  Top 10 Habitable Planets
       │
       ▼
  Feature Preprocessing  (utils.py)
       │
       ▼
  exohabit_model.pkl  ──►  Prediction Output

---

🟢 Endpoint 1 — Health Check

GET / — Confirms the backend API is running.

// Response
{
  "message": "ExoHabitAI Backend API Running",
  "status": "success"
}

↑ GET 127.0.0.1:5000 — API health confirmed live in browser

---

🔵 Endpoint 2 — Model Information

GET /model-info — Returns model metadata and features used.

// Response
{
  "features_used": [
    "planet_radius",
    "planet_mass",
    "orbital_period"
  ],
  "model_name": "ExoHabitAI",
  "version": "1.0"
}

↑ GET 127.0.0.1:5000/model-info — model metadata returned

---

🚀 Endpoint 3 — Habitability Prediction

POST /predict — Predicts habitability of an exoplanet from input features.

# Tested via backend/test_api.py
python backend/test_api.py

// Response
{
  "habitability_probability": 0.9995276927947998,
  "prediction": 1,
  "status": "success"
}

prediction: 1 → Habitable · prediction: 0 → Non-Habitable

↑ test_api.py output — 99.95% habitability probability returned

---

🏆 Endpoint 4 — Habitability Ranking

GET /rank — Returns the Top 10 most habitable exoplanets from the dataset, ranked by habitability probability and supporting metrics including ESI score, equilibrium temperature, orbital stability factor, and stellar compatibility index.

// Response (truncated)
{
  "count": 10,
  "status": "success",
  "top_planets": [
    {
      "Equilibrium_Temp": 753.2,
      "Habitability_Probability": 0.9999881,
      "Habitability_Score_Index": 0.2260783639999819,
      "Orbital_Period": 18.2734,
      "Orbital_Stability_Factor": 0.40665921750518514,
      "Planet_Density": 3.53,
      "Planet_Mass": 8.7,
      "Planet_Name": "TOI-134564 b",
      "Planet_Radius": 2.469,
      "Stellar_Compatibility_Index": 0.8883289840950909,
      "Stellar_Metallicity": 0.0,
      "Stellar_Temp": 5048.0,
      "Target_Habitable": 1
    }
    // ... 9 more planets
  ]
}

↑ GET /rank — Top 10 habitable exoplanet candidates with full metrics

---

🧪 Features at a Glance

Feature	Description
🔭 Habitability Prediction	Binary classification — prediction: 1 (Habitable) or 0 (Non-Habitable) via POST /predict
🌍 ESI Calculation	Earth Similarity Index computed from planetary radius, density, escape velocity, and surface temperature
🤖 Multiple ML Models	Logistic Regression, Decision Tree, Random Forest, XGBoost — best model selected and deployed
📊 Rich Visualizations	Confusion matrices, ROC curves, feature importance charts, missing-value heatmaps
🖥️ Interactive Dashboard	Space-themed frontend with real-time prediction UI at frontend/dashboard.html
🟢 Health Check API	GET / — confirms backend is live
🔵 Model Info API	GET /model-info — returns model name, version, and features used
🚀 Prediction API	POST /predict — returns habitability_probability (0–1) and binary prediction
🏆 Ranking API	GET /rank — returns Top 10 habitable exoplanet candidates with full metrics
📦 Deployment Ready	Configured for Render via render.yaml — live at onrender.com

---

📂 Project Structure

B13-EXOHABITAI/
│
├── 📁 assets/images/        # README screenshots & visuals
├── 📁 backend/              # Flask API
│   ├── app_deploy.py        # 4 endpoints: /, /model-info, /predict, /rank
│   ├── utils.py             # Feature preprocessing
│   └── best_model.pkl       # Serialized production model
│
├── 📁 data/
│   ├── raw/                 # NASA Exoplanet Archive snapshot (CSV)
│   ├── preprocessed/        # Cleaned & imputed dataset
│   └── processed/           # Feature-engineered + habitability-ranked output
│
├── 📁 frontend/
│   ├── index.html           # Landing page
│   └── dashboard.html       # Prediction UI
│
├── 📁 models/
│   └── exohabit_model.pkl   # Tuned Random Forest — final production model
│
├── 📁 notebooks/
│   ├── DataPreprocessingCodeFile.ipynb
│   ├── model_training.ipynb
│   └── ExoHabitAI_Project_Report.pdf
│
├── 📁 reports/              # Model evaluation outputs (19 PNGs + 2 CSVs)
│   ├── final_model_metrics.csv
│   ├── model_comparison.png
│   ├── feature_importance.png
│   └── [confusion matrices & ROC curves for all 5 models]
│
├── render.yaml              # Render deployment config
├── requirements.txt
└── README.md                # You are here 📍

↑ Full project directory — VS Code Explorer

---

🛠️ Tech Stack

Layer	Technology
Language	Python 3.10+
ML Framework	Scikit-Learn, XGBoost
Data Processing	Pandas, NumPy
Visualization	Matplotlib, Seaborn
Backend / API	Flask
Frontend	HTML5, CSS3, JavaScript
Serialization	Pickle (.pkl)
Deployment	Render (render.yaml)
Notebooks	Jupyter
Version Control	Git & GitHub

---

🚀 Run It Locally — Launch Sequence

Prerequisites

Python 3.10+
pip
Git

Step 1 — Clone the Repository

git clone https://github.com/YOUR_USERNAME/B13-ExoHabitAI.git
cd B13-ExoHabitAI

Step 2 — Create Virtual Environment

python -m venv venv
# Windows
venv\Scripts\activate
# macOS/Linux
source venv/bin/activate

Step 3 — Install Dependencies

pip install -r requirements.txt

Step 4 — Launch the Flask API

cd backend
python app_deploy.py

Step 5 — Open the Dashboard

Navigate to: http://localhost:5000

🟢 You're live. Enter exoplanet parameters and receive your first habitability prediction.

---

☁️ Deployment

ExoHabitAI is configured for Render deployment via render.yaml.

Deploy to Render

1. Fork this repository
2. Connect to Render and create a New Web Service
3. Point it to your forked repo
4. Render auto-detects render.yaml and configures the build
5. Set environment variables if needed
6. Click Deploy 🚀

---

📈 Future Improvements

The mission doesn't end here — it's just beginning

• 🔭 Real-time NASA API Integration — pull live exoplanet data
• 🧠 Deep Learning Models — LSTM/Transformer-based habitability scoring
• 🌐 3D Planet Visualization — interactive Three.js exoplanet renderer
• 📱 Mobile-Responsive UI — full cross-device support
• 🔍 Explainability Layer — SHAP values for every prediction
• 🗃️ Database Integration — store and compare historical predictions
• 🛰️ Multi-Star System Support — binary star habitability zones

---

🤝 Contributing

Contributions, ideas, and pull requests are welcome.

# Fork → Branch → Commit → PR
git checkout -b feature/your-feature-name
git commit -m "feat: describe your change"
git push origin feature/your-feature-name

Please follow clean commit conventions and add docstrings to new functions.

---

📄 License

This project is licensed under the terms of the LICENSE file included in this repository.

---

---

🌌 "We are made of star stuff. And somewhere out there, another world might be too."

---

ExoHabitAI was built at the intersection of data science and wonder.
Every model trained, every feature engineered, every prediction made —
is one small step toward answering humanity's oldest question:

Are we alone?

Made with 🔭 by Samridhi Gupta — exploring worlds beyond Earth, one dataset at a time.