Data analysis depends on quality of input data but this increase chance of privacy break of organization or individual or community. So reverse mining process is applied that performs both the data privacy preserving and knowledge extraction. In order to improve education quality student data analysis is more sensitive and needs good set of features for prediction. This paper has proposed a model that extracts features from the different city schools and trains a model for grade prediction. Proposed model has not shared student data to any third party, instead of this random features selected by the Wolf Optimization genetic algorithm were used for the training of model. These features were taken in form of presence and absence of student activities. Experiment was done on real dataset of Maharashtra Districts School Students. Comparisons result shows that proposed model has improved the prediction accuracy by % as compared to similar models of privacy preserving

Data mining, Genetic Algorithm, Privacy Preserving, Neural Network