A PCA-based variable ranking and selection approach for electric energy load forecasting

Abstract

Purpose This paper aims to propose an approach based upon the principal component analysis (PCA) to define a contribution rate for each variable and then select the main variables as inputs to a neural network for energy load forecasting in the region southeastern Brazil. Design/methodology/approach The proposed approach defines a contribution rate of each variable as a weighted sum of the inner product between the variable and each principal component. So, the contribution rate is used for selecting the most important features of 27 variables and 6,815 electricity data for a multilayer perceptron network backpropagation prediction model. Several tests, starting from the most significant variable as input, and adding the next most significant variable and so on, are accomplished to predict energy load (GWh). The Kaiser–Meyer–Olkin and Bartlett sphericity tests were used to verify the overall consistency of the data for factor analysis. Findings Although energy load forecasting is an area for which databases with tens or hundreds of variables are available, the approach could select only six variables that contribute more than 85% for the model. While the contribution rates of the variables of the plants, plus energy exchange added, have only 14.14% of contribution, the variable the stored energy has a contribution rate of 26.31% being fundamental for the prediction accuracy. Originality/value Besides improving the forecasting accuracy and providing a faster predictor, the proposed PCA-based approach for calculating the contribution rate of input variables providing a better understanding of the underlying process that generated the data, which is fundamental to the Brazilian reality due to the accentuated climatic and economic variations.

Impacts of Feature Selection on Predicting Machine Failures by Machine Learning Algorithms

“…With respect to attribute selection, although the works of Giordano et al [10], Chang et al [11], and Bezerra et al [19] have addressed industrial applications, none of them address the application of attribute selection techniques to prioritize the selection of the most relevant sensors for data collection.…”

“…Subsequently, PCA identifies additional orthogonal directions, each capturing subsequent levels of variability, creating subsequent principal components. This iterative process continues until the principal components encapsulate all significant variations in the data [18][19][20].…”

“…Performs dimensionality reduction by projecting the data into a new feature space composed of the most significant principal components [18][19][20].…”

“…In this work, PCA was applied to the data as indicated by [19,20] to verify each variable's percentage of total variance using the main components. To do this, we employed Sklearn's PCA Decomposition library in Python, which has a function called ex-plained_variance_ratio.…”

Traceless Kalman filter threshold estimation for distributed power loads

“…There are many current modeling methods for power load forecasting, and power load forecasting models with good performance have emerged [3]. Traditional models include models such as linear regression and moving averages, which collect and analyze the historical data of power loads, from which they find the changing law between power loads and factors and fit the future value of power loads according to the changing law [4][5][6]. The traditional model assumes that the power load is a fixed and unchanging trend, such as an upward trend or a downward trend [7][8][9].…”