Modeling Uncertainties in Electricity Consumption Using the Z-Score Approach

The pursuit of energy efficiency in power systems emphasizes the need to accurately model uncertainties in electricity consumption, which are inherently stochastic due to human behavior. Non-parametric methods offer precise representations but may become impractical when data is limited. Conversely, parametric methods, such as the Z-Score, provide more straightforward solutions by assuming a known distribution, although they may incompletely capture data variability. This study evaluates the applicability of the Z-Score for modeling consumption uncertainties by comparing it to equivalent percentiles using real-world data. Error metrics such as MAE, MSE, RMSE, and SMAPE are used to quantify the differences between the approaches. Although exact equivalence between Z-Scores and percentiles is uncommon, the results highlight that the parametric method provides accurate estimates in intermediate ranges, making it a viable option for modeling uncertainties in electricity consumption under data-limited scenarios.