Privacy preserving in data mining - Experimental research on SMEs data

Analysis of data on individuals and business sensitive data as well as revealing the results of such analysis without disclosing confidential and sensitive information is a very important issue. Many techniques for preserving privacy of data are currently being used. This paper is addressing some of the basic techniques: randomization, k-anonymity, distributed privacy preserving and application effectiveness downgrading. Most of the techniques should be applied in the phase of data collection or their preprocessing, which can lead to different results (better or worse) of data mining than would be obtained on original data. For this reason, data analysts should be encouraged to quantify the ratio between privacy preserved in data with application of each technique and the loss of data or quality of outputs. This paper illustrates the application of certain techniques for preserving privacy on experimental dataset, and reveals the effects that their use has on the results.