YGGDRASIL - A Statistical Package for Learning Split Models

TL;DR

This paper introduces a statistical framework for context-specific independence models called split models and presents YGGDRASIL, a software package for learning these models from data.

Contribution

It develops a comprehensive framework for split models and provides a software tool for their statistical inference and learning.

Findings

Framework for context-specific independence structures

Software package YGGDRASIL for model learning

Enhanced modeling capabilities over traditional graphical models

Abstract

There are two main objectives of this paper. The first is to present a statistical framework for models with context specific independence structures, i.e., conditional independences holding only for sepcific values of the conditioning variables. This framework is constituted by the class of split models. Split models are extension of graphical models for contigency tables and allow for a more sophisticiated modelling than graphical models. The treatment of split models include estimation, representation and a Markov property for reading off those independencies holding in a specific context. The second objective is to present a software package named YGGDRASIL which is designed for statistical inference in split models, i.e., for learning such models on the basis of data.