Comparative study for software effort estimation by soft computing models

Abstract EN

Accurate estimation of software development effort has become a crucial for effective projects planning.

It is a very challenging task for project teams to predict the development effort required in the initial phases of a software project.

Software estimation prior to development process can decrease the risk and enhance the project's success rate.

Although numerous traditional and machine learning models have been proposed for software effort estimation over the past decade, the level of accuracy is not satisfactory enough.

The objective of this study is to assess the capability of using two different soft computing methods namely artificial neural networks (ANNs) and adaptive neuro-fuzzy inference system (ANFIS) to estimate the software effort.

COCOMO dataset is used to test and train the proposed models.

Mean magnitude of relative-error (MMRE) and correlation coefficient (R) were used as assessment criteria.

It was concluded that both models can satisfactorily estimate software development efforts, but ANFIS model has outperformed the ANN model in two statistical indicators: MMRE and correlation R.

It is recommended that ANFIS can be used as a predictive model for software effort estimation.