Application of Nonparametric Methods in Studying Energy Consumption

Abstract

Consumer behaviour towards different forms of energy varies over time. The variance can be so large that the quality of the estimation functional relationship between the response variable and its associated explanatory variables is seriously affected. To attenuate this, kernel smoothing a nonparametric regression approach is proposed. This approach offers a powerful tool in modelling and adapts to various types of designs. The aim of this study is to produce a reasonable model that defines the structural change of a stationary time series which exhibits volatility over time. The explanatory variable used is the lagged values of the series. To study the effects at the tails, the quantiles are proposed. This model is functional in examining the characteristics of peak hour electricity consumption in Kenya. It is found that the mean peak consumption is a decreasing function of the lagged time and that the more extreme the peak consumption, the higher the volatility. This model provides insights on routine shift time energy consumption modelling