Article Preview
TopIntroduction
The EU has set binding targets of a 40% reduction of domestic greenhouse gas emissions (compared to 1990 levels) to be reached by 2030 and a share of renewable energy of 32% (EC, 2018). The ability to accurately forecast electricity consumption is of great importance for utility providers in order to predict future costs, enhance the financial bottom line and reduce negative environmental impacts. Utility providers try to achieve these goals by using the ‘smart grid’ and ‘demand-side management’ (Corbett, 2013). Demand-side management is focused on the downstream activities related to the consumption-end of the value chain, with the objective of understanding, influencing, and managing consumer demand (Canever et al., 2008). Through incentives or different electricity prices, demand response activities motivate behavior change (Albadi and El-Saadany, 2008). Smart meters are increasingly becoming common, and they are expected to foster intelligent energy consumption behavior through the flow of information between users and utility providers.
There are IS-enabled information processing capacities within smart meters that have a significant impact on the effectiveness of demand-side management (Corbett, 2013). Most of the demand-side research has focused on adoption and outcome of the adoption of smart meters by individuals (see Chou et al., 2017; Kuo et al., 2018; Murray et al., 2018; Dehdarian 2018; Hackbarth and Löbbe, 2018; Wunderlich et al., 2019; Hielscher and Sovacool, 2019). Previous studies also have been predominantly restricted to factors that can be managed and manipulated by utilities but are silent on the factors specific to residential and household characteristics (Corbett et al., 2018). Wunderlich et al. (2019) argue few studies in IS literature have examined ‘household’ technologies in general and smart meters in particular, resulting in gaps in our understanding of why and how households adopt such novel and often complex technologies leading to calls for more research on this topic at the household level (Venkatesh et al., 2016).
‘Energy pricing’ and ‘environmental concerns’ have been identified as the key influencing factors with regards to energy consumption behavior (Karjalainen 2011; Vassileva et al., 2012c). However, the persistence of these factors over time and the impact of other important factors have not been fully investigated by previous studies (Loock et al., 2013). Hence, the objectives of the current study are twofold. The first objective is to segment households based on socio-demographic factors and energy consumption behavior. Identifying different clusters of households is central to understanding the consumption behavior and the household’s motivations for energy consumption reduction and participation in smart meter trial. Different clusters of households bring out specific contextual nuances that could develop our understanding of consumption behavior. The second objective is to investigate the impact of smart meter installation on attitude change towards energy consumption behavior. Previous research has found motivation and attitude towards conserving energy are pivotal in saving energy (Oltra et al., 2013; Hackbarth and Löbbe, 2018) so it is essential to have a deep understanding of how attitudes and behaviors towards energy consumption are formed and changed in order to develop more successful interventions by policymakers.
The utilization of data mining techniques to model residential electricity consumption has so far been limited in information systems (IS) literature. Few studies in the past have utilized data mining segmentation techniques to identify factors influencing residential energy consumption behavior (Baker and Rylatt, 2008; Van Raaj and Verhallen, 1983). A couple of other studies have also used data mining classification techniques, namely, decision trees and neural networks (Yu et al., 2010).
Moreover, previous research has mainly used low-resolution and aggregated energy consumption data due to lack of advanced metering technologies. We use electricity consumption data obtained via the smart meters and combine with socio-economic and behavioral data to understand different groups of electricity consumers. For customer segmentation, it is crucial to obtain information on the ‘actual energy consumption’ of customers using smart meters rather than low resolution and aggregated data, which is mainly used by previous research.