This program:

Data Collection: Fetches historical transaction fee data from Blockchain.com API
Feature Engineering: Creates time-based features and rolling averages
Machine Learning: Uses Random Forest regression to predict fees
Visualization: Plots historical fee data
Prediction: Estimates fees for specific times and categorizes them as Low/Medium/High

Key features:

Uses multiple time-based features (hour, day of week, etc.)
Includes 24-hour rolling average for context
Categorizes predictions based on historical percentiles
Provides visual representation of fee patterns
Calculates model accuracy using MAE

To use:

Install required packages: pip install pandas numpy scikit-learn matplotlib requests
Run the script
The program will:

    Download 60 days of historical data

    Train a prediction model

    Show a fee history chart

    Predict fees for 6 hours in the future