Page 22 - Contributed Paper Session (CPS) - Volume 2

CPS1408 Caston S. et al.

Probabilistic hourly load forecasting
2
1
1
* Caston Sigauke , Murendeni Nemukula , Daniel Maposa 3
1 Department of Statistics, University of Venda Private Bag X5050, Thohoyandou 0950, South
Africa
2 School of Mathematics, Statistics and Computer Science University of KwaZulu-Natal, South
Africa
3 Department of Statistics and Operations Research University of Limpopo, South Africa

Abstract
The paper discusses short-term hourly load forecasting using additive quantile
regression (AQR) models. A comparative analysis is done using generalised
additive models (GAMs). In both modelling frameworks, variable selection is
done using least absolute shrinkage and selection operator (Lasso) via
hierarchical interactions. The AQR model with pairwise interactions was found
to be the best fitting model. The forecasts from the models are then combined
using a convex combination model and also using quantile regression
averaging (QRA). The AQR model with interactions is then compared with the
convex combination and QRA models and the QRA model gives the most
accurate forecasts. The QRA model has the smallest prediction interval
normalised average width and prediction interval normalised average
deviation. The modelling framework discussed in this paper has established
that going beyond summary performance statistics in forecasting has merit as
it gives more insight into the developed forecasting models

Keywords
Additive quantile regression; Lasso; load forecasting; generalised additive
models

1. Introduction
Recent work on short-term load forecasting in which hourly data is
modelled separately includes that of ([3, 4, 5, 11]). Goude et al. [5] developed
generalised additive models for forecasting electricity demand. The authors
used hourly load data from 2260 substations across France. Individual models
for each of the 24 hours of the day were developed. The developed models
produced accurate forecasts for the the short and medium term horizons.
Additive quantile regression models for forecasting probabilistic load and

1
* Corresponding author: Tel.: +27 15 962 8135 Email address: csigauke@gmail.com (C.
Sigauke).

11 | I S I W S C 2 0 1 9