Analysis of clustering and classification algorithms to forecast electric power consumption

Abstract

Our era is characterized by the wide spread of data of all kinds to the extent that it has become impossible for analysts to extract meaningful information by resorting only to traditional approaches to preliminary data analysis. With the presence of large amounts of stored data, the need has increased to develop tools that are characterized by strength and speed to analyze it, and extract information and knowledge from it. Hence, data mining techniques emerged as techniques aimed at extracting knowledge from huge amounts of data (known in turn as Big Data) [2].Forecasting electric energy consumption requires knowledge of daily consumption quantities, consumption times and other influencing factors that constitute large amounts of data that can be analyzed using data mining algorithms. The accurate prediction of electrical load is still a challenging task due to many problems such as the non-linear nature of the time series or the seasonal patterns it displays, which are very time consuming and also affect the accuracy of the prediction performance. The process can be improved by using the support beam (SVM) algorithm and other algorithms. Initially, the climatic factors, type, location and period of consumption and a number of other influencing factors were studied in order to improve the performance of the electric power consumption forecasting process.The following clustering and classification algorithms were also analyzed to predict the electric power consumption, where it was suggested to use the SVM algorithm to solve the electric load prediction problem, which provided a solution to the classification and regression problems and also helped classify the categorical data and gave an independent optimization solution of the model compared to previous studies. A group of classifiers such as random forests, support vector machine, neural networks, and others were also used to reach accurate results that help to take appropriate decisions in the field of electricity consumption prediction. In the last stage, based on the prediction values ​​resulting from this study, work was done to distribute electrical energy in the most appropriate manner and in line with the importance of higher usage so that we have the ability to operate energy sources at specific times and in appropriate quantities to try to reduce the waste caused by operating unnecessary sources.