Determining the Risky Software Projects using Artificial Neural Networks

Determining risky software projects early is a very important factor for project success. In this study it aimed to choose the most correctly resulting modelling method that will be useful prediction of help companies avoid losing time and money on unsuccessful also facing legal requirements because not being able fullfill their responsibilites customers While making research subject, seen in previous researches, usually traditional techniques were preferred. But observed these methods mostly resulted with high misclassification ratio. To overcome problem, proposes three-layered neural network (NN) architecture backpropagation algorithm. NN was trained by using two different data sets which OMRON set (collected OMRON) 2016-2020 ES.LV authors) separately. For made firstly relevant classification (Gaussian Naive Bayes Algorithm) (Scaled Conjugate Gradient Backpropagation chosen both each seperately purpose observing type would give better results. Experimental results showed approach predicting whether or risky.