A Statistical Comparative Study For Detecting Outliers In Electrical Data

The rapid development of artificial intelligence models in electrical engineering drew attention to the outlier detection methods because they represent one of the main causes of model overfitting. In this paper, three popular outlier detection algorithms, boxplots, z-score and standard deviation, are analyzed in the context of electrical data composed of voltages, currents and harmonics. The study aims to establish an algorithm that can accurately detect outliers so the data can be used in others more advanced models.