You have a dataset of patient records for diabetes testing, including the patient's age. You plan to generate a report with the mean age calculated from differentially private data derived from this dataset.

To minimize the risk of exposing the actual data, which epsilon value should you use for the analysis?