k-diagram: Polar Insights for Forecasting

Navigate the complexities of forecast uncertainty and model behavior with specialized polar visualizations.

Welcome to the official documentation for k-diagram. This package provides a unique perspective on evaluating forecasting models, especially when uncertainty quantification is crucial. Dive in to discover how polar plots can reveal deeper insights into your model’s performance, stability, and potential weaknesses.

Install k-diagram

Quick Start Guide

Plot Gallery

Key Features

• Intuitive Polar Perspective: Visualize multi-dimensional aspects like uncertainty spread, temporal drift, and spatial patterns in a compact circular layout.

• Advanced Error Analysis: A dedicated suite of plots to diagnose systemic bias vs. random error, compare error distributions, and visualize 2D uncertainty.

• Targeted Diagnostics: Functions specifically designed to assess interval coverage, consistency, anomaly magnitude, model velocity, and drift.

• Uncertainty-Aware Evaluation: Move beyond point-forecast accuracy and evaluate the reliability of your model’s uncertainty estimates.

• Identify Model Weaknesses: Pinpoint where and when your forecasts are less reliable or exhibit significant anomalies.

• Clear Communication: Generate publication-ready plots to effectively communicate model performance and uncertainty characteristics.

See Also

Quick links to the main sections of the API Reference:

• Uncertainty Visualization: Functions for analyzing prediction intervals, coverage, anomalies, and drift.

• Error Visualization: Functions for diagnosing systemic bias, variance, and comparing error distributions.

• Model Evaluation: Functions for generating Taylor Diagrams to compare model performance.

• Feature Importance: Functions for visualizing feature influence patterns (fingerprints).

• Relationship Visualization: Functions for plotting true vs. predicted values in polar coordinates.

• Full API Reference: Browse the complete API documentation.