interpylate-fls 0.0.0

A Python package for training neural networks on 5-dimensional datasets for interpolation tasks

Interpylate-FLS

A Python package for training neural networks on 5 dimensional datasets for interpolation tasks.

Features

• NeuralNetwork: PyTorch-based neural network for regression tasks
• DataLoader: Dataset loading and preprocessing utilities
• Plotter: Visualization tools for model performance
• Logger: Logging utilities

Installation

PYPI

pip install interpylate-fls

Local Installation

pip install -e .

Example Usage

from interpylate_fls import NeuralNetwork, DataLoader

# Load and preprocess data
loader = DataLoader('data.pkl')
data = loader.load_dataset()
X_train, X_test, X_val, y_train, y_test, y_val, _, _, _ = loader.inspect_data(data)

# Create and train neural network
nn = NeuralNetwork(
    X_train=X_train,
    y_train=y_train,
    X_val=X_val,
    y_val=y_val,
    X_test=X_test,
    y_test=y_test,
    hidden_layer_sizes=[64, 32],
    learning_rate=0.001,
    epochs=100
)

# Train the model
nn.train(verbose=True)

# Evaluate
mse, r2 = nn.evaluate()
print(f"MSE: {mse:.4f}, R²: {r2:.4f}")

# Make predictions
prediction = nn.predict([0.5, 0.5, 0.5, 0.5, 0.5])
print(f"Prediction: {prediction}")

Requirements

• Python >= 3.9
• PyTorch >= 2.2.0
• scikit-learn >= 1.6.0
• pandas >= 2.3.0
• numpy >= 1.26.0, < 2.0.0
• matplotlib >= 3.9.0
• seaborn >= 0.13.0

License

MIT License

Author

Funmi Looi-Somoye