SYNOPSIS
Sensitivity analysis is carried out to validate model formulation. A modified model has been developed to predict the future energy requirement of coal, oil and electricity considering price, income, technological and environmental factors. The impact and sensitivity of the independent variables on the dependent variable are analysed. The error distribution pattern in the modified model as compared to a conventional time series model indicated the absence of clusters. The residual plot of the modified model showed no distinct pattern of variation. The percentage variation of error in the conventional time series model for coal and oil ranges from −20% to + 20%, while for electricity it ranges from −80% to + 20%. However, in the case of the modified model the percentage variation in error is greatly reduced—for coal it ranges from −0.25% to + 0.15%, for oil −0.6% to + 0.6% and for electricity it ranges from −10% to + 10%. The upper and lower limit consumption levels at 95% confidence is determined. The consumption at varying percentage changes in price and population are analysed. The gap between the modified model predictions at varying percentage changes in price and population over the years from 1990 to 2001 is found to be increasing. This is because of the increasing rate of energy consumption over the years and also the confidence level decreases as the projection is made far into the future.