Evaluation of Discretization Effect Towards Numerical Data for Data Protection

We evaluated a method to protect individual privacy. Existing privacy methods create a trade-off between confidentiality and openness. To balance these, we evaluated the effect of discretization on protecting confidential numerical data. We evaluated a method that uses discretization by frequency or binning and adjusted the discretization parameter. By adjusting the number of bins, we showed that anonymized data is 30% harder to de-anonymize during data reconstruction attacks.