Abstrait

Multiple Class Identification and Categorization with Nuggets on Health Care Data

P. Jamuna, G, Mohana Prabha, M.E, (Ph.D)

Finding the patterns and outliers is one of the major problems in the field of data mining. Especially in the field of health care analysis has become difficult to predict the patterns and decision making. Classification techniques are used to identify the transaction label. The classification techniques are used to collect the patterns in the learning phase and detect the outliers in training phase. In health care analysis, only classifications are limited with two class levels as positive and negatives. The symptoms of patients are collected and categorized into patterns then by using the patterns; they detect the severity level of diseases. The proposed system mainly focuses on detecting the severity level of patients by enhancing the boundary classifications. This idea can be achieved by critical nuggets which is a record or attribute used to define classification where that attribute considered as the deciding authority. The classification accuracy can be improved with critical nuggets and enhancing to support multi class (low, medium, high and normal) and multiple attribute environment.The critical nuggets identification and classification scheme is improved to support multiple classes. The system can be adopted to handle mixed attribute data values. The boundary approximation algorithm is enhanced to reduce the detection complexity. Post processing operations are tuned to identify classes for multiple category data environment.