The Uganda Electricity Transmission Company (UETC) has the monopoly for buying bulk electricity from generating companies within Uganda and selling it to local distributors and other distributors within the East African region. However, up to now, monthly units of electricity sales are reported in tables or graphs and forecasts are mainly judgmental. This limitation fails to provide the quantitative input data required for management planning and orecasting based on sales.
The purpose of this study was to identify the main characteristics in the trends of the monthly sales, analyse the sales using simple linear regression and Auto Regressive Integrated Moving Average (ARIMA)
models to determine a good fit estimate model, which was thereafter evaluated in an interactive analysis. The study used Eviews 3.1 econometric package covering the period February 2006 to July 2011.
It revealed a general upward trend in sales with variations from one month to another without any recognisable seasonality. A simple linear regression estimate model was found to be unsuitable for representing the sales since most actual values were outside one standard confidence
limits of the estimated model. However, using the ARIMA model estimate and analysis through identifying and carrying out predictive and diagnostic tests, a good fit was ARIMA(1,1,0), represented by the equation Ye = 0.9675 + Y(t-1) - 0.6004Y(t-1) + 0.6004Y(t-2) showing that its estimate depended mainly on the last two most recent lag monthly sales after first order differencing. Furthermore, an evaluation forecast showed that the actual sales fitted in the one standard confidence limits 2 Freddie Festo Mawanga ORSEA Journal of the model, thus confirming that the model could be used for forecasting sales beyond July 2011.
This study provides managers of UETC with a practical and yet simple tool to describe, explain, intervene in and orecast monthly electricity sales that are required by management for planning.
The purpose of this study was to identify the main characteristics in the trends of the monthly sales, analyse the sales using simple linear regression and Auto Regressive Integrated Moving Average (ARIMA)
models to determine a good fit estimate model, which was thereafter evaluated in an interactive analysis. The study used Eviews 3.1 econometric package covering the period February 2006 to July 2011.
It revealed a general upward trend in sales with variations from one month to another without any recognisable seasonality. A simple linear regression estimate model was found to be unsuitable for representing the sales since most actual values were outside one standard confidence
limits of the estimated model. However, using the ARIMA model estimate and analysis through identifying and carrying out predictive and diagnostic tests, a good fit was ARIMA(1,1,0), represented by the equation Ye = 0.9675 + Y(t-1) - 0.6004Y(t-1) + 0.6004Y(t-2) showing that its estimate depended mainly on the last two most recent lag monthly sales after first order differencing. Furthermore, an evaluation forecast showed that the actual sales fitted in the one standard confidence limits 2 Freddie Festo Mawanga ORSEA Journal of the model, thus confirming that the model could be used for forecasting sales beyond July 2011.
This study provides managers of UETC with a practical and yet simple tool to describe, explain, intervene in and orecast monthly electricity sales that are required by management for planning.
Table of Contents
Copyright © 2012. All Rights Reserved.
Site developed & maintained by ICTC, UoN