Quality of the software is an important factor for any software company. Software fault prediction is a data mining process that helps to improve the quality. Data mining tools both open source and proprietary are available today. These bring lots of research works in this area. Software fault is the bug in the software that is identified only after its installation and it makes the software behave not in the expected way. Bug is there even after testing due to various constraints like cost, time. Prediction will help identify those fault prone areas and with that one can concentrate on those modules in future. Hybrid Feature Selection and Hybrid Classifier approach is a way to improve the software fault prediction accuracy. In Hybrid feature selection, irrelevant, redundant features are first filtered and this filtered feature set reduces the input feature set of wrapper. In Hybrid Classifier approach Linear Discriminant Analysis score is used as an additional feature for Neural Network classifier. These models give a better fault prediction accuracy.