Arch Technologies — Internship Project 2
Predicting Bitcoin's close price using historical market data with XGBoost and an interactive Gradio app.
This project predicts Bitcoin’s closing price based on historical trading data.
It was developed during my internship at Arch Technologies and focuses on:
- Financial data preprocessing
- Feature engineering
- Machine learning model training
- Model performance evaluation
- Interactive deployment with Gradio
- Load and preprocess raw cryptocurrency data
- Select relevant features for price prediction
- Train a robust XGBoost Regressor model
- Evaluate results using MAE & R² metrics
- Visualize predictions vs actual prices
- Deploy an interactive web app for real-time predictions
| Category | Tools / Libraries |
|---|---|
| Data Processing | pandas, numpy |
| Visualization | matplotlib, seaborn |
| Machine Learning | scikit-learn, xgboost |
| Deployment | gradio |
| File Handling | openpyxl |
File: btc_data.xlsx
Features used for prediction:
open— Opening pricehigh— Highest pricelow— Lowest pricevolume— Trading volumemarketCap— Market capitalization
Target variable:
close— Closing price
git clone https://github.com/yourusername/btc-price-prediction.git
cd btc-price-prediction