AN IMPROVED BAT ALGORITHM FOR SOFTWARE COST ESTIMATION

摘要

We optimize the cost estimation of the COnstructive COst Model, COCOMO, by using the improved bat algorithm. Thebat algorithm depicts the hunting and routing behavior of bats which is based on their natural echolocation facility. This algorithmperforms the optimization by switching from exploration to exploitation automatically. Furthermore, the performance of thealgorithm is improved by introducing a random walk in the exploration phase which speeds up convergence towards global optima.The improved algorithm determines the optimized value of the constants used in the COCOMO model to get a more accuratesoftware cost estimation. The algorithm is tested on four datasets downloaded from the data repository and compared with existingstate of the art techniques. The result proves the significance of the technique.