A bank is building a machine learning model to predict loan approvals. Before training the model, the data science team wants to check whether the historical training dataset already contains imbalances across gender and age groups that could lead to unfair outcomes. Which AWS capability best supports this specific need?