|
---
|
|
title: "๐ Gemstone Price Regression"
|
|
emoji: ๐ฐ
|
|
colorFrom: indigo
|
|
colorTo: blue
|
|
sdk: streamlit
|
|
app_file: app.py
|
|
pinned: true
|
|
license: mit
|
|
tags:
|
|
- regression
|
|
- machine-learning
|
|
- streamlit
|
|
- diamonds
|
|
- synthetic-data
|
|
---
|
|
|
|
# ๐ Gemstone Price Prediction App
|
|
|
|
This Streamlit app predicts the price of a gemstone using its physical and quality-related features.
|
|
|
|
## ๐ง Project Overview
|
|
|
|
- This project simulates a **gemstone pricing system** using synthetic tabular data.
|
|
- Features include: `carat`, `depth`, `table`, `x`, `y`, `z`, `clarity_score`, `color_score`, and `cut_score`.
|
|
- The target variable is **price** (USD).
|
|
- Model: **RandomForestRegressor**
|
|
- Trained on 1000 synthetic samples.
|
|
|
|
---
|
|
|
|
## ๐ Performance
|
|
|
|
- RMSE: **605.16**
|
|
- Rยฒ Score: **0.9549**
|
|
|
|
---
|
|
|
|
## ๐ How to Run Locally
|
|
|
|
```bash
|
|
pip install -r requirements.txt
|
|
streamlit run app.py
|
|
|
|
|
|
|
|
๐ฎ Future Work
|
|
Area Improvement
|
|
Model Try XGBoost, LightGBM
|
|
Feature Engineering Interaction terms, log/carat scaling
|
|
Deployment Add API endpoint with FastAPI
|
|
Real-world Data Integrate real gemstone datasets
|
|
|
|
|
|
๐ Files
|
|
app.py: Streamlit interface
|
|
|
|
rf_model.pkl: Trained model
|
|
|
|
model_columns.pkl: List of input features
|
|
|
|
requirements.txt: Required libraries
|
|
|
|
|