machine learning apps
Exoplanet Type Classifier
An interactive PyTorch and FastAPI application that predicts exoplanet types from planetary, orbital, and host-star features.
PythonPyTorchFastAPIScikit-learnPandasNumPyHTMLCSSJavaScript
Project Overview
Exoplanet Type Classifier is an interactive multiclass classification web application that predicts the type of an exoplanet using planetary, orbital, and host-star features. The project combines a PyTorch neural network model with a FastAPI backend and a custom browser-based interface.
Project Goal
- Build a multiclass classifier for exoplanet type prediction.
- Create an interactive interface where users can adjust planetary and host-star features.
- Return predicted class probabilities and a readable planet-type profile.
- Deploy the FastAPI application as a live portfolio demo.
Model Architecture
1.A PyTorch MLP model performs multiclass classification over exoplanet type categories.
2.FastAPI serves the frontend, health endpoint, UI configuration, and prediction endpoint.
3.The browser interface uses sliders and preset cards to simulate different exoplanet profiles.
4.The prediction result displays confidence, class probabilities, and typical profile information for the predicted planet type.