In your role as a Machine Learning Engineer, you're working on a classification problem involving time series data. After minimal experimentation using random cross-validation, you've achieved an unusually high 99% AUC ROC on the training data, without employing sophisticated algorithms or extensive hyperparameter tuning. Given the potential issues this could indicate, such as data leakage or overfitting, and considering the constraints of ensuring model generalizability and compliance with data privacy standards, what should be your next step to accurately identify and resolve the underlying issue? Choose the best two options.