phamluan/crypto-polkadot-predictor

Polkadot (DOT) Price Prediction Models

Trained ML models for predicting Polkadot (DOT) cryptocurrency prices.

📊 Model Performance

Model	RMSE	MAE
Random Forest	0.2348	0.1403
Gradient Boosting	0.2031	0.1157
Linear Regression	0.0200	0.0149
LSTM	0.2332	0.1725

🎯 Training Details

• Trained on: 2025-10-24 07:45:22
• Data Source: CoinGecko API
• Historical Days: 365
• Features: 23 technical indicators
• GPU: Accelerated with TensorFlow

📦 Files Included

• polkadot_sklearn_models.pkl: Scikit-learn models (RF, GB, LR)
• polkadot_scaler.pkl: Feature scaler
• polkadot_lstm_model.h5: LSTM neural network
• polkadot_metadata.json: Training metadata

🚀 Usage

from huggingface_hub import hf_hub_download
import joblib
from tensorflow.keras.models import load_model

# Download models
sklearn_path = hf_hub_download(
    repo_id="YOUR_USERNAME/YOUR_REPO",
    filename="polkadot_sklearn_models.pkl"
)
scaler_path = hf_hub_download(
    repo_id="YOUR_USERNAME/YOUR_REPO",
    filename="polkadot_scaler.pkl"
)
lstm_path = hf_hub_download(
    repo_id="YOUR_USERNAME/YOUR_REPO",
    filename="polkadot_lstm_model.h5"
)

# Load models
models = joblib.load(sklearn_path)
scaler = joblib.load(scaler_path)
lstm = load_model(lstm_path)

# Make predictions
# (prepare your features first)
predictions = models['RandomForest'].predict(scaled_features)

📈 Features

The models use 23 technical indicators including:

• Moving Averages (SMA 7, 25, 99)
• Exponential Moving Averages (EMA 12, 26)
• RSI (Relative Strength Index)
• MACD & Signal Line
• Bollinger Bands
• Stochastic Oscillator
• Volatility measures
• Lag features

⚠️ Disclaimer

These models are for educational and research purposes only. Cryptocurrency markets are highly volatile and unpredictable. Do not use these predictions for actual trading decisions without proper risk management.

📄 License

MIT License