Privacy preserving data mining approach for extracting fuzzy rules

Privacy preserving data mining is an active area of research. In today´s world, with the growth of technology, huge amounts of data are continuously being gathered and kept in databases. Indeed, temporal associations and correlations among items in large transactional database that help in many business decision-making processes. The main aim of privacy preserving data mining is concerned with developing the data mining techniques which is applied on datasets without violating the privacy of individuals. The proposed work deals with extracting fuzzy rules using privacy preserving data mining. The dataset used in this work is Indian Liver Patient Dataset. The liver data is preprocessed in order to fill the missing values. The first level of privacy preserving is done using data transformation technique. Data transformation is the process of converting data or information from one format to another. The second level of privacy preserving is done using fuzzy logic. This level is required to handle the uncertainty in the dataset and it is the process of transforming a scalar value into fuzzy values. The fuzzified values are used to extract fuzzy rules. The rule extraction is done using neural network algorithm. To extract the rules, initially the neural network is trained and pruned using the single hidden layer. Then the hidden activation values is generated by using logistic function. The decision tree is used to classify the rules.