Privacy-preserving mining by rotational data transformation

Many data mining applications deal with large data sets that contain private information that must be protected. This has led to the development of many privacy-preserving data mining techniques. Many of these techniques use randomized data distortion by adding noise to the sensitive data. However, non-careful noise addition may introduce biases to the statistical parameters of these data, including means and variances. To meet privacy requirements and preserve the statistical properties of the sensitive data we use a data transformation technique called Rotation-Based Transformation (RBT). This method distorts only confidential numerical attributes and preserves the statistical properties of the data.