Model Persistence

Guide for saving and loading trained forecasting models.

Overview

The model persistence functionality enables you to: - Save trained forecasters to disk - Load previously trained models - Manage model caching - Handle batch model operations

Functions

API Reference

Full API documentation for all model persistence functions (save_forecasters, load_forecasters, ensure_model_dir, model_directory_exists) is auto-generated in the Processing Reference.

Saving Models

See save_forecasters in the API reference.

Loading Models

See load_forecasters in the API reference.

Model Directory Management

See ensure_model_dir and model_directory_exists in the API reference.

Examples

from spotforecast2_safe.manager.persistence import (
    save_forecasters,
    load_forecasters,
)

# Save trained models
trained_forecasters = {...}  # Your trained forecasters
save_forecasters(trained_forecasters, model_dir="models/")

# Load previously trained models
forecasters, missing = load_forecasters(
    target_columns=["power"], 
    model_dir="models/"
)