Chen, Q, Fan, Z, Kaleshi, D and Armour, S (2015) Rule Induction-Based Knowledge Discovery for Energy Efficiency. IEEE ACCESS, 3. 1423 - 1436.

[img]
Preview
Text
07219372.pdf - Published Version
Available under License Creative Commons Attribution Non-commercial.

Download (4MB) | Preview

Abstract

Rule induction is a practical approach to knowledge discovery. Provided that a problem is developed, rule induction is able to return the knowledge that addresses the goal of this problem as if-then rules. The primary goals of knowledge discovery are for prediction and description. The rule format knowledge representation is easily understandable so as to enable users to make decisions. This paper presents the potential of rule induction for energy efficiency. In particular, three rule induction techniques are applied to derive knowledge from a dataset of thousands of Irish electricity customers’ time-series power consumption records, socio-demographic details, and other information, in order to address the following four problems: 1) discovering mathematically interesting knowledge that could be found useful; 2) estimating power consumption features for customers, so that personalized tariffs can be assigned; 3) targeting a subgroup of customers with high potential for peak demand shifting; and 4) identifying customer attitudes that dominate energy conservation.

Item Type: Article
Additional Information: This is the final published version of the article (version of record). It first appeared online via IEEE at http://doi.org/10.1109/ACCESS.2015.2472355 - please refer to any applicable terms of use of the publisher.
Uncontrolled Keywords: Energy efficiency, knowledge discovery, smart grids, subgroup discovery
Subjects: Q Science > QA Mathematics > QA75 Electronic computers. Computer science
Divisions: Faculty of Natural Sciences > School of Computing and Maths
Related URLs:
Depositing User: Symplectic
Date Deposited: 11 May 2018 07:46
Last Modified: 01 Apr 2019 14:00
URI: http://eprints.keele.ac.uk/id/eprint/4880

Actions (login required)

View Item View Item