Support Vector Machine Regression Predicts Energy Consumption and Conservation Attitude in Households

Abstract

The purpose of this study is twofold. First, causal mechanisms underlying the effect of changing attitudes on behavioral change are investigated. Second, recommendations are presented to improve the understanding of how attitudes and behavioral patterns of energy consumption in the residential sector can change. To that end, a theoretical model is developed and tested with empirical data. Also systematically investigates the association strength of each input variable with each of the output variables using various classical statistical analysis tools to identify the most strongly related input variables. Thereafter, three learning approaches, namely, multiple linear regression, polynomial linear regression, and support vector regression, are applied to predict the changing attitude and pattern of behavior variables. Results show that the performance of SVR with kernel radial basis function and polynomial regression hat of other forecasting models. However, the significant nonlinearity between inputs and outputs should be further developed to improve forecast precision. The study shows that cognitive factors are the most decisive factor in behavioral patterns and that the behavioral approach is strongly affected. Moreover, motivations and cognitive factors were found to have the most substantial effect on changing patterns of behaviors.