README.md

---
title: StarSystemClassification
emoji: 🚀
colorFrom: red
colorTo: red
sdk: docker
app_port: 8501
tags:
- streamlit
pinned: false
short_description: A Streamlit app that predicts of a star system
license: mit
---

# 🪐 Star System Classification (LightGBM)

This project predicts the **system_type** of a star system using astrophysical and galactic features.
It is a **multiclass classification** task with 4 classes (0–3).

## ✅ What this app does
- Takes 10 input features (numeric + categorical)
- Applies the **same preprocessing** used in training:
  - `stellar_activity_class` mapped to numbers (Low/Medium/High)
  - `planet_configuration` and `star_spectral_class` encoded using saved `LabelEncoder`s
  - Features are ordered using the saved `feature_order` file
- Predicts the star system type using a **LightGBM** model

## 📦 Files in this repository
Required files (must be in the same folder as `app.py`):

- `app.py`
- `lightgbm_model.pkl`  (saved LightGBM model)
- `planet_encoder.pkl`  (LabelEncoder for `planet_configuration`)
- `star_encoder.pkl`    (LabelEncoder for `star_spectral_class`)
- `featurer.pkl`        (saved feature order list)
- `requirements.txt`

## 🚀 Run locally
```bash
pip install -r requirements.txt
streamlit run app.py