API Reference

This section provides comprehensive documentation for all DRN classes, functions, and modules. The API documentation is automatically generated from the source code docstrings, similar to R package documentation.

Overview

The DRN package is organized into several key modules:

• Models - Core distributional regression models (GLM, DRN, CANN, MDN, DDR)
• Distributions - Custom distribution implementations and utilities
• Training - Training functions, loss functions, and optimization utilities
• Metrics - Evaluation metrics for distributional forecasting
• Interpretability - Model explanation and visualization tools
• Utilities - Data preprocessing, splitting, and helper functions

Quick Reference

Core Classes

Class	Purpose	Key Methods
BaseModel	Abstract base for all models	.fit(), .predict(), .quantiles()
GLM	Generalized Linear Models	.fit(), .predict(), .clone()
DRN	Distributional Refinement Network	.fit(), .predict(), .log_adjustments()

Key Functions

Function	Purpose	Module
train()	Train models with PyTorch	drn.train
drn_loss()	DRN loss function	drn.models
crps()	Continuous Ranked Probability Score	drn.metrics
split_and_preprocess()	Data preprocessing	drn.utils