Abstract:

The evolution of smart homes is very rapid and the benefits, comfort, as well as flexibility in controlling energy consumption, attract the development smart home culture across the globe. The energy consumption data collected from these smart homes play a major role in energy pricing, understanding consumers’ behavior, demand-side management, etc., functionalities. But, sometimes, this collected data may suffer from the anomalies such as missing data, redundancy, outliers, etc., which affect the energy data analytics. Among these anomalies, the missing data is one of the anomalies to be concentrated more as it makes data incomplete and significantly hinders the further analysis of the data. This missing of data may take place in three different patterns viz. missing completely at random, missing at random, and missing not at random. Therefore, capturing the pattern of the missing data is highly preferred to better handle them. Although there are a few works on the missing data, they are focused only on the occurrence, behavior, impacts, recovery, and imputation of the missing data rather than identifying the pattern of missing data. Hence to address this problem, this paper proposes a statistical approach to ascertain the pattern of missing data in the energy consumption data of smart homes. The proposed statistical approach revealed that the data are missing at random in the energy consumption data. An energy consumption database named ‘Tracebase’ is used for implementing the proposed approach.

Keywords: