While developing software it is very important that the software should be of defect free. But, none of the software can be 100% defect free and various studies are in progress to build a model which minimizes the defect as much as possible by predicting it at an early stage of development. Based on the probability facts various researchers has used probabilistic model to predict defects in the program. To contribute in this research and enhancing the existing model of software defect prediction we are proposing a model based on the combination of probabilistic and deterministic model through defect association learning. The experimental evaluation in comparison with the existing methods shows the improvement in the accuracy of predicting the defect by using Deterministic and Probabilistic defect prediction.