Expert system means the system which makes decision based on provided knowledge by human expert and displays the result. But, knowledge provided by the human expert may varies from other expert. So, sometimes this procedure becomes subjective to the human expert. To eliminate this, we use decision tree classifier for initial phase of the expert system. From the decision tree we select the minimum list of attributes from the initial dataset and create new minimal dataset for use in Inference engine. In second phase we use naive bays classifier to create the prediction model. Which improve the accuracy of prediction of class on testing dataset.