Analytical model for residential predicting energy consumption

![CDATA[Effective energy consumption prediction is important for determining the demand and supply of energy. The challenge is how to predict energy consumption? This study presents an energy consumption analytical regression model and process based on the project conducted in an Australian company. This study involved the analysis of household and energy consumption datasets in the residential sector. The analytical model generation process is organised into four major stages: prepared the household and energy consumption data or data cleansing, household energy consumption clustering (segmentation or groups) using k-means clustering algorithm for similarity measure in their characteristics, stepwise multiple regression for variables selection to determine the final model's predictors, and filter the final regression model to identify the influential observations using Cook's distance and Q-Q (quantile-quantile) normal plot for improvement in the model. The final filtered regression model represents 64 percent variation to the dependent variable is explained by independent variables with correlation 0.8 between energy consumption observed and predicted values. The abovementioned process and resultant regression model seem useful for developing household energy consumptions models for managing the demand and supply of energy.]]