Effects of Data Standardization on Hyperparameter Optimization with the Grid Search Algorithm Based on Deep Learning: A Case Study of Electric Load Forecasting

This study investigates data standardization methods based on the grid search (GS) algorithm for energy load forecasting, including zero-mean, min-max, max, decimal, sigmoid, softmax, median, and robust, to determine hyperparameters of deep learning (DL) models. The considered DL models are convolutional neural network (CNN) long short-term memory (LSTMN). procedure is made over (i) setting configuration CNN LSTMN, (ii) establishing hyperparameter values LSTMN epoch, batch, optimizer, dropout, filters, kernel, (iii) using eight standardize input data, (iv) GS optimal mean absolute error (MAE) percent (MAPE) indexes. effectiveness proposed method verified power Australian state Queensland Vietnamese Ho Chi Minh city. simulation results show that appropriate, except zero-mean min-max methods.