Modeling risk prediction of diabetes — A preventive measure

Databases in clinical scenario have tremendous amount of data regarding patients and clinical history associated. Here, data mining plays vital role in searching for patterns within huge clinical data that could provide useful basis of knowledge for efficient and effective decision-making. Classification mechanism is widely used tool of data mining employed in healthcare applications to facilitate disease diagnosis and prediction. Usually medical dataset are high dimension in nature containing many insignificant attributes or features and result poor classification with inaccuracies. Feature selection is a technique used for preprocessing the high-dimensional data to reduce data dimension and to remove redundant and irrelevant features. This paper provides a systematic data mining approach for selecting best indicators of diabetes among many attributes present in the database and gives an appropriate model to track the diabetes before its onset. It selects the most appropriate classifier model for the given dataset through voting mechanism to achieve best accuracy and eliminating any biased result.