NASA Space Apps Challenge 2025

A World Away: Hunting for Exoplanets with AI

Team: Minima acción

Challenge: Hunting for Exoplanets with AI

Project Overview

This repository contains a reproducible Machine Learning solution designed to detect exoplanets using the Kepler Object of Interest (KOI) dataset. The system utilizes AutoGluon for automated model selection and training, wrapped in a Streamlit web interface that allows users to perform real-time predictions and retrain models dynamically.

Key Features

• AutoML Integration: Utilizes AutoGluon to benchmark and select the best-performing models for exoplanet classification.
• Interactive Dashboard: A Streamlit-based UI (Front/) to visualize data and interact with the model.
• Dynamic Retraining: Capabilities to update the model with new datasets directly from the interface.

---

Repository Structure

The project is organized into backend logic (ML pipeline) and frontend presentation (Web App).

.
├── Back/
│   ├── AutogluonModels/             # Serialized trained models
│   └── Exoplanet-Detection-with-ML.ipynb  # Core training notebook (Jupyter)
├── Data/
│   ├── 1_cumulative_2025.csv        # Primary Dataset (KOI cumulative)
│   ├── 2_TOI_2025.csv               # TESS Objects of Interest
│   └── 3_k2pandc_2025.csv           # K2 Candidates
├── Front/
│   ├── app_streamlit_min.py         # Main entry point for the Web App
│   ├── home.py                      # Landing page logic
│   ├── pagina_modelo.py             # UI module for model retraining
│   ├── pagina_predictor.py          # UI module for inference/prediction
│   ├── create_test_from_csv.py      # Data transformation pipeline
│   └── static/                      # Assets (Team photos, Logos, SHAP plots)
├── Buscando Exoplanetas.pdf         # Project presentation/Paper
└── requirements.txt                 # Dependencies

Prediction Interface

Model Analysis (SHAP)

Installation & Deployment

Prerequisites

• Linux Environment
• Python 3.9+
• 8 GB RAM (Minimum for AutoGluon training)
• Virtual Environment support

Setup Steps

1. Clone the repository:

git clone git@github.com:JSR-Mario/NASA.git
cd NASA

2. Environment Setup: Create and activate a virtual environment to isolate dependencies.

python3 -m venv .venv
source .venv/bin/activate

3. Install Dependencies:

pip install -r requirements.txt

4. Launch the Application: Navigate to the frontend directory and start the Streamlit server.

cd Front
streamlit run app_streamlit_min.py

*Access the UI at: http://localhost:8501*

---