Hybrid Model to Improve Bat Algorithm Performance Solution for Software Cost Optimization

Bat Algorithm is one of the successful metaheuristic algorithms, which is used prominently for the purpose of optimization. But its inherent feature of non-changing parameters with the various iterations makes it less appropriate for optimization of software cost estimation techniques like COCOMO. So the current study proposes a hybrid model for the improvement of Bat algorithm by enhancing the search (global) and thus helping in optimizing the fitness function by generating new solutions. The data set used for testing is NASA 63 and the fitness function used for cost estimation is Mean Magnitude of Relative Error (MMRE). The simulations are done using MATLAB version R2010a. Results shows a better MMRE for the hybrid model as compared to the original Bat algorithm used for the optimization of COCOMO II for software cost estimation.