Research of Applying Simulated Annealing to Software Effort Estimation

• Main Authors: Tso-Chia Chang, 張祚佳
• Other Authors: Sun-Jen Huang
• Format: Others
• Language: zh-TW
• Published: 2005
• Online Access: http://ndltd.ncl.edu.tw/handle/72598648422115561033

碩士 === 國立臺灣科技大學 === 資訊管理系 === 93 === Accurate estimate of software effort is a key to achieving successful development of a software project. Parametric model with nonlinear function is a commonly adopted software development effort estimation method that has been evaluated and confirmed in past two decades. Hartley’s modified Gauss-Newton method is the most widely used approach on computing nonlinear least squares estimators in many nonlinear parametric models. A nonlinear parametric model with suitable initial values plays an important role as a poor choice of starting values may result in convergence to a local minimum. Simulated Annealing (SA) method is a global search algorithm to solve the optimization problems and can prevent local optimization for exploring suitable values. In this paper, we aim at investigating the effect of improving software effort estimate accuracy by applying SA to parametric-based software effort estimation models. SA is utilized to determine the suitable valuesof paramenters in parametric-based software effort estimation models with different software effort estimation functions and software cost drivers. The experimental result shows that applying SA to build parametric software effort estimation model can produce a more accurate estimate than the model of traditional nonlinear regression method.