Knowledge Discovery and Data Mining
Presents many fundamental technologies and recent advances in data mining and knowledge discovery.
TABLE OF CONTENTS Knowledge Discovery and Data Mining
Chapter 12: Data Mining with Neural Networks An Applied Example in Understanding Electricity Consumption Patterns
Overview
Philip Brierley and Bill Batty
Cranfield University, UK
Data mining is the process of analysing data in order to extract useful information. Many techniques exist that can be used in the analysis of data, one of which is artificial neural networks.
In this chapter we explain what a neural network is and how one was used in analysing electricity consumption data from a utility in the UK. Total daily loads over an eight-year period were examined and the influencing factors identified. Rules were subsequently extracted from the neural network by pruning weights in order to ease comprehension of how the data was being processed. A half-hourly model for a single year was created and more subtle factors which influence the load in this timescale were identified.
The ethos in this work is to utilise the modelling capabilities of neural networks in order to identify anomalies to the model. The network results merely provide clues as to why such anomalies might exist, but it is the human ability to interpret the results which is the actual knowledge-discovery process. The neural network is simply employed as a data-mining tool that can model large amounts of data but it has no intelligence for knowledge discovery whatsoever. It is hoped that the approach taken will demonstrate that neural networks can be very powerful tools for analysing large amounts of data, if they are understood and taken for what they are.
The appendixes contain a proof of the backpropagation weight update rule and code...
Copyright The Institution of Electrical Engineers 1999 under license agreement with Books24x7
