An Accurate Effort Estimation using Fuzzy Based Alternating Regression Technique (FBART)

摘要

Precise software effort estimation is considered as the most significant task in the software development life cycle. Since, software developers use effort estimate as an input for planning budgets that highlights pricing policies and investment analysis. The over estimation or underestimation of software effort can result in detrimental effect on the quality of software and may cause huge monetary loss. Hence, there arises a need for developing reliable software effort estimation techniques that uses software effort impact factors. The Least Effort Multiplier based Fuzzy Estimation Algorithm (LEMFEA) [1] is an enhanced version of FFPA-PSR (fuzzy-based function point analysis with performance metrics, security, and reliability factors) [2] algorithm that has been proposed for improving the accuracy of the software effort estimation. In LEMFEA uses fuzzy logic to frame rules based on the classification of the attributes like project type (T), programmers skill (S), software language used (L), database used (D) and criticality (C) required for the estimation. In the proposed method Fuzzy Based Alternating Regression Technique (FBART) alternates between two regression techniques, viz., i) Gauss-Newton based model and ii) Weighted Deming Model based on the size of the project estimated based on function points and lines of code. This FBART is an adaptation of LEMFEA that has been proposed for improving the accuracy of the software effort estimation. This FBART algorithm uses Re-usability (R) as an extra multiplier for effective software effort estimation. The experiments are carried out and the performance of the proposed software estimation techniques are validated and the results confirms that Fuzzy Based Alternating Regression Technique (FBART) is found better to the existing software estimation techniques.