NovaTrace

NovaTrace is a web application for detecting and classifying exoplanets using a machine learning model trained on NASA datasets. Users can upload their own data, explore predictions, retrain the model, and visualize planetary systems in 3D.

Demo video in less than 30 seconds: NovaTrace

---

Table of Contents

• Project Setup
• Backend
• Frontend
• Frontend Tabs
  • Upload
  • Classification
  • Data Preview
  • 3D System Visualization
  • Data Exploration

---

Project Setup

1. Clone the repository or download and extract zip file:

git clone https://github.com/JonathanSolvesProblems/NovaTrace.git

2. Navigate to the backend folder and set up the backend server, as per steps below.
3. Navigate to the frontend folder and set up the frontend application, as per steps below.

Backend

The backend handles communication with the machine learning model. The model is trained on the following NASA datasets:

• Kepler Objects of Interest (KOI): Link
• TESS Objects of Interest (TOI): Link
• K2 Planets and Candidates: Link

Steps to run backend locally:

cd backend
python -m venv venv
source venv/bin/activate   # On Windows: venv\Scripts\activate
pip install -r requirements.txt
uvicorn server:app --reload --port 8000

The backend will start at http://localhost:8000 and handle requests from the frontend.

Frontend

The frontend provides a user interface for uploading data, viewing classifications, visualizing systems, and exploring the dataset.

Steps to run frontend locally:

cd frontend
npm install
npm run dev

The frontend will be available at http://localhost:5173 (or the port displayed in the console).

Frontend Tabs

Upload

The Upload tab is the home page where users can upload Excel files containing exoplanet candidate data. Uploaded data is processed and sent to the model for classification.

Optional: The Retrain Model tab allows users to retrain the machine learning model using their uploaded data. Hyperparameters can be adjusted, though this step is optional. Retraining helps the model adapt to new datasets.

Classification

The Classification tab displays the model's predictions. Example:

Classification Results Example

• CONFIRMED: High confidence exoplanet detections (50%)
• CANDIDATE: Potential exoplanets requiring verification (50%)
• FALSE POSITIVE: Non-planetary signals (0%)

CONFIRMED Objects Example

Object ID	Period (days)	Radius (R⊕)	Stellar Temp (K)	Classification	Confidence
Object-1	—	—	—	CONFIRMED	62%
Object-2	—	—	—	CONFIRMED	71%

---

Data Preview

The Data Preview tab shows uploaded data in a table with search functionality.

Example Columns:

• koi_period
• koi_duration
• koi_depth
• koi_prad
• koi_teq
• koi_insol
• koi_model_snr
• koi_steff
• koi_srad
• koi_disposition

Users can search and filter data to inspect individual records.

---

3D System Visualization

The 3D System Visualization tab provides an interactive 3D view of detected exoplanets:

• 🖱️ Drag to rotate
• 🔍 Scroll to zoom
• ⚡ Real-time orbital motion

Displays detected planets, host star type, system architecture, orbital dynamics, discovery method, and habitability status.

---

Data Exploration

The Data Exploration tab provides visual analysis of model predictions and astrophysical features:

• Overview
• Scatter Plots
• Histograms

Classification Distribution Example:

• 2 Candidates
• 2 Confirmed

Key Insights Example:

• 2 classification categories detected
• Typical exoplanet radii cluster around median values
• Stellar temperatures range across the dataset
• Explore scatter plots to see clustering by orbital period and planet radius