There has been a tremendous growth in the demand for software fault prediction during recent years. In this paper, Levenberg-Marquardt (LM) algorithm based neural network tool is used for the prediction of software defects at an early stage of the software development life cycle. It helps to minimize the cost of testing which minimizes the cost of the project. The methods, metrics and datasets are used to find the fault proneness of the software. The study used data collected from the PROMISE repository of empirical software engineering data. This dataset uses the CK (Chidamber and Kemerer) OO (object-oriented) metrics. The accuracy of Levenberg-Marquardt (LM) algorithm based neural network are comparing with the polynomial function-based neural network predictors for detection of software defects. Our results indicate that the prediction model has a high accuracy. General TermsSoftware defect prediction

General TermsParticle Swarm Optimization, Neural network, Bayesian Network.

No abstract

No abstract