Parameter Estimation of COCOMO II using Simulated Annealing

Software cost estimation is process of forecasting the effort required to build up a software engineering project. This process becomes one of the principal challenges and most expensive component in the field of software. The precise results are must for proper project planning because any error may result in huge losses. The cost estimation is usually reliant upon the size estimate of the project. There are a number of diverse techniques for executing software cost estimation amid which COCOMO II is commonly used because of its lucidity and simplicity. Effort estimation models are based on various soft computing techniques such as neural network, tabu search, genetic algorithm, fuzzy logic modeling etc. for finding the precise predictive software development effort and time estimation. But the author intends to use the Simulated Annealing approach to handle these models and will show potential advantages in solving the problem. This technique will minimize functions of various variables. This technique will be applied to arbitrary combinatorial problems. The COCOMO II model predicts software development effort in Person Months (PM) and project duration in months. This work aims to propose simulated annealing for optimizing current coefficients of COCOMO II model to achieve more accuracy in estimation of software development effort.