Chapter 7. Learning Models

Table of Contents

Overview
The Modeling Process in Proximity
Relational Bayesian Classifier
Code example: run-1d-clusters-rbc.py
Relational Probability Trees
Code example: run-1d-clusters-rpt.py
Understanding and viewing relational probability trees
Using temporal attributes in a relational probability tree
Relational Dependency Networks
Code example: run-1d-clusters-rdn.py
Viewing relational dependency networks
Tips and Reminders

Overview

The exercises in this chapter walk through the process of training, testing, and evaluating several probabilistic models designed for relational knowledge discovery. For each exercise in this chapter, we first walk through the source code and identify how Proximity classes and methods are used to accomplish the desired task. This is followed by instructions for executing the script in Proximity.

The examples in this chapter are written in Jython, a Java implementation of Python that lets you interact with Java code. The classes and methods used in these examples are, of course, also available for use in Java code. Source code files for all the scripts discussed in this chapter are available in $PROX_HOME/doc/user/tutorial/examples.

Objectives

The exercises in this chapter demonstrate how to

  • learn, apply, and evaluate the relational Bayesian classifier model

  • learn, apply, and evaluate the relational probability tree model

  • learn, apply, and evaluate the relational dependency network model

  • use temporal attributes to restrict the set of applicable attribute values in a relational probability tree

  • display and interpret graphical representations of relational probability trees

  • display and interpret graphical representations of relational dependency networks