Diabetes is the most common chronic disease among
the world. Early prediction of these will assist the physicians to
provide the improved treatment. Machine learning approaches
are widely used for predicting the disease at the earlier stage.
However the selecting the significant features and the suitable
classifier are still reduces the diagnosis accuracy. In this paper
the PCA based feature transformation and the hybrid random
forest classifier is utilized for diabetes prediction. PCA attempt to
identify the best subset of transformed components that greatly
improves the classification result. The system is compared with
priori machine learning approaches to evaluate the efficiency of
this work. The experimental result shows that the present study
enhances the prediction accuracy.