International Journal of Computer & Organization Trends(IJCOT)
Privacy preservation is vital for machine learning and data mining, but measures created to protect financial information sometimes bring about a trade off: reduced utility of the workout samples. This paper introduces a privacy preserving approach that could be put on decision-tree learning, without decrease in accuracy. It describes a procedure for the preservation of privacy of collected data samples if information of one's sample database continues to be partially lost. Existing approach will not work well for sample datasets with low frequency, or if low variance within the distribution of every samples.