The use of hybrid analytics to establish effective strategies for household energy conservation

Abstract

Taiwan highly relies on foreign countries' supply in various types of energy because of the shortage of energy resources. Taiwanese government encourages industrial and academic research organizations to invest research in energy conservation that could be applicable to countries with similar settings. This research analyzed energy conservation and demographic data of households in Taiwan and developed a novel method for best setting household energy conservation strategies. Through the analysis of high volume and high dimensional data, it showed a hybrid analytical method is required, to predict and identify target households and refine their energy conservation strategies. The hybrid method includes a Logistic Regression based predictive model and a Rough Set Theory-based decision-making model. A hybrid analytical approach is used to identify target households and the critical energy conservation factors by households, such as the use of air conditioning in terms of time length and time of the day, age of the house, energy conservation of the electrical appliances, and price promotion. References show that no study was conducted with such a hybrid approach that can refine strategies for the households that were false-positive in achieving the conservation goal. Besides the above-mentioned purpose, this novel method intended to lower the promotion cost, as promotion is a part of the strategy refinement. In overall, the following problems were considered for decision-making: 1. how to identify the target households that require strategy refinement, 2. how to identify key variables and determine the change of these variables to increase the likelihood of achieving conservation goal, 3. what options can be provided to the decision-makers for feasible strategy refinements. Some studies with qualitative approaches and judgmental decision-making methods were found in this area. However, with the available large dataset, the proposed quantitative method achieved better decision-making efficiency and result optimality.