A Machine Learning-powered application that predicts the 22K gold price per gram (INR) in India based on the USD to INR exchange rate.
🚀 Built as a job-ready ML project to demonstrate skills in data scraping, preprocessing, model building, evaluation, and web deployment using Gradio.
Gold prices in India fluctuate daily and are strongly influenced by the USD to INR exchange rate. This project:
- Collects historical gold prices (22K per gram) and USD/INR rates.
- Trains a regression model to predict daily gold price.
- Deploys the prediction interface using Gradio for real-time input/output.
- Gold Price Data: Scraped using
requestsandre(fromexchange-rates.org) - USD to INR Data: Downloaded via
yfinance
| Category | Tools / Libraries |
|---|---|
| Language | Python |
| Data Collection | requests, re, yfinance, pandas |
| Visualization | matplotlib, seaborn, plotly.express |
| Data Scaling | StandardScaler |
| ML Models | LinearRegression, RandomForestRegressor |
| Model Selection | train_test_split |
| Evaluation Metrics | MAE, MSE, R² Score |
| Interface / UI | Gradio |
| Serialization | pickle |
- 📈 Predicts 22K gold price per gram in INR.
- 🏦 User inputs the current USD to INR exchange rate.
- 🧠 Trained using Random Forest Regressor for higher accuracy.
- 🌍 Deployed using Gradio for interactive web access.
| Metric | Score |
|---|---|
| MAE | 262.26 |
| MSE | 186,641.77 |
| R² Score | 0.8563 |
🎯 Achieved strong accuracy with just one feature (USD to INR), demonstrating practical modeling skills and potential for further enhancement.
Try it out:
🔗 https://huggingface.co/spaces/sankarans2001/gold-price-predictor
📁 gold-price-predictor/ ├── gold_dataset.csv # Web-scraped dataset ├── scaler.pkl # Pre-fitted StandardScaler ├── rfr_model.pkl # Trained RandomForestRegressor ├── app.py # Gradio interface logic ├── gold-prediction.pdf # End-to-end notebook pdf: data → model → evaluation ├── README.md # Project documentation