An Improved Cost Estimation in Software Project Development Using Neural Networks and COCOMOII model

An sympathetic of quality aspects is relevant for the software association to deliver high software dependability. An empirical consideration of metrics to prophesy the quality attributes is basic in order to acquire insight about the value of software in the primitive phases of software development and to certify corrective actions. Herein paper, we forecast a model to assess fault proneness via Object Oriented CK metrics and QMOOD metrics. We pertain one statistical method and six machine learning technique to predict the models. The proposed reproduction are validated using dataset unruffled from Open Source software. The consequences are analyzed using Area Under the Curve (AUC) achieve from Receiver Operating Characteristics (ROC) testing. The results show that the replica predicted using the random forest and bagging methods outperformed all the other mould. Hence, support on these results it is equitable to claim that quality models have a considerable relevance with Object Oriented metrics and that machine learning organizations have a equivalent performance with numerical methods. It is experimental that the CBR routine using the Mahalanobis detachment similarity occupation moreover the inverse distance weighted solution algorithm yielded the best fault prediction. In addition, the CBR models have superior performance than models basis on multiple linear regression.