Abstract

Contents Page

Code Description

Introduction

We want to create a personal learning assistant to schedule and help ease the burden of scheduling daily events by learning the user’s environment [2]. Here we look at decision trees, in particular ID3 and Quinlan’s C4.5, expecting more from ID3 based on previous work with ID3 and a hierarchal ID3 by Hsu and Lin [3].

Since the personal assistant is continually learning from data and trying to predict values, we consider learning sets of rules. In Learn-One-Rule, a general to specific search, as describe in Mitchell, [5], is based on the ID3 algorithm. Hence, ID3 is used here to learn the sets of rules. Since ID3 is the first method chosen and the tree was generated using WEKA software, the personal assistant’s learner easily makes use of other decision trees output by WEKA. Quinlan’s C4.5 is an interesting choice since it has many option of using unpruned trees and pruned trees.