This study explores the impact of context on the explanation of regression-based models, particularly in scenarios involving categorical features. It has been established that when using Shapley values to explain regression models, accounting for the context in which the model operates is essential for providing accurate interpretations. In this work, we introduce the concept of adjusting the context through the creation of mean backgrounds, a technique previously unstudied in the context of categorical features. By modifying the “Mean Background Method” to accommodate categorical variables, we present a novel approach to controlling context when calculating Shapley values. The proposed method was validated through its application to a model predicting electrical load, demonstrating its ability to effectively handle categorical features. Additionally, the study examines the influence of various parameters of the modified Mean Background Method, shedding light on its robustness and versatility. These findings contribute to the broader understanding of model explainability and offer a more comprehensive approach to Shapley-based explanations in regression settings.