The author introduced an efficient security mechanism during the classification of data. The author proposed a cryptographic algorithms that protects centralized data sets for decision tree data mining. Cryptographic Technique is applied to the sample data sets priority to their release to third parties in order to threat of their inattentive disclosure or theft. Their approach does not affect the accurate of data mining results. The decision tree can be constructed from the data sets, such that the originals data sets do not need to be reformatted.