Analysis of Business Customers’ Energy Consumption Data Registered by Trading Companies in Poland
Open Access
Abstract
In this article, we analyze the energy consumption data of business customers registered by trading companies in Poland. We focus on estimating missing data in hourly series, as forecasts of this frequency are needed to determine the volume of electricity orders on the power exchange or the contract market. Our goal is to identify an appropriate method of imputation missing data for this type of data. Trading companies expect a specific solution, so we use a procedure that allows to choose the imputation method, which will consequently improve the accuracy of forecasting energy consumption. Using this procedure, a statistical analysis of the occurrence of missing values is performed. Then, three techniques for generating missing data are selected (missing data are generated in randomly selected series without missing values). The selected imputation methods are tested and the best method is chosen based on MAE and MAPE errors.Keywords
Funding Information
- European Regional Development Fund (13/010/BK_21/0057)
- National Center for Research and Development (13/010/BK_21/0057)
- Silesian University of Technology (13/010/BK_21/0057)
This publication has 43 references indexed in Scilit:
- Forecasting electricity smart meter data using conditional kernel density estimationOmega, 2016
- A New Missing Data Imputation Algorithm Applied to Electrical Data LoggersSensors, 2015
- Missing value imputation strategies for metabolomics dataElectrophoresis, 2015
- Imputation of missing data in time series for air pollutantsAtmospheric Environment, 2015
- Missing value imputation in high-dimensional phenomic data: imputable or not, and how?BMC Bioinformatics, 2014
- Solar radiation forecast based on fuzzy logic and neural networksRenewable Energy, 2013
- A 24-h forecast of solar irradiance using artificial neural network: Application for performance prediction of a grid-connected PV plant at Trieste, ItalySolar Energy, 2010
- Pattern classification with missing data: a reviewNeural Computing and Applications, 2009
- Imputation of missing values is superior to complete case analysis and the missing-indicator method in multivariable diagnostic research: A clinical exampleJournal of Clinical Epidemiology, 2006
- Software cost estimation with incomplete dataIEEE Transactions on Software Engineering, 2001