Improving Software Effort Estimation Using Neuro-Fuzzy Model with SEER-SEM

Accurate software development effort estimation is a critical part of software projects. Effective development of software is based on accurate effort estimation. Although many techniques and algorithmic models have been developed and implemented by practitioners, accurate software development effort prediction is still a challenging endeavor in the field of software engineering, especially in handling uncertain and imprecise inputs and collinear characteristics. In order to address these issues, previous researchers developed and evaluated a novel soft computing framework. The aims of our research are to evaluate the prediction performance of the proposed neuro-fuzzy model with System Evaluation and Estimation of Resource Software Estimation Model (SEERSEM) in software estimation practices and to apply the proposed architecture that combines the neuro-fuzzy technique with different algorithmic models. In this paper, an approach combining the neuro-fuzzy technique and the SEER-SEM effort estimation algorithm is described. This proposed model possesses positive characteristics such as learning ability, decreased sensitivity, effective generalization, and knowledge integration for introducing the neuro-fuzzy technique. Moreover, continuous rating values and linguistic values can be inputs of the proposed model for avoiding the large estimation deviation among similar projects. The performance of the proposed model is accessed by designing and conducting evaluation with published projects and industrial data. The evaluation results indicate that estimation with our proposed neuro-fuzzy model containing SEER-SEM is improved in comparison with the estimation results that only use SEERSEM algorithm. At the same time, the results of this research also demonstrate that the general neuro-fuzzy framework can function with various algorithmic models for improving the performance of software effort estimation.