Optimization of COCOMOII Model Coefficients using Tabu Search

Software effort estimation is one of the most important step in project planning. Accurate estimates help the company in decision making process and effectively manage the software development process. Various Software cost estimation models have been proposed till now, COCOMO81 is one of them. It became one of the most popular parametric cost estimation model in 1980s but when new life cycle processes came it experienced some difficulties in calculating cost. COCOMOII model came in 1995 to overcome these problems and became one of the most commonly used model for estimating effort. Todays models are based on various soft computing techniques such as fuzzy logic, neural network, genetic algorithm, simulated annealing, tabu search etc. Fuzzy logic and neural network techniques are very hard to use and genetic algorithm cannot produce very good results because of its problem of premature convergence. This work aims to propose Tabu Search algorithm which can increase the efficiency of output values of parameters of COCOMOII model and can evaluate predicted effort values as accurate to the real values.