<p>This study proposes a counterfactual explanation method to obtain fair results for sensitive attributes that can be applied in various situations, including education. Two methods are proposed: one uses the average as fairness, and the other derives fair results among sensitive attributes. The proposed methods are applied to simulation and actual educational data. Machine learning can be used to model complex data from simulations using nonlinear models. Moreover, the methods were able to identify the crucial variables for achieving fairness and predicting outcomes after implementing the changes. An analysis of the real data was performed to extract the variables that are important for fair results. For instance, “group study and group work” was important for academic achievement. These methods can be applied in several cases where fair results are important.</p>