An Approach to Software Cost Estimation by Improved — Time Variant Acceleration Coefficient Based PSO
Prince Bhatia, KK Mishra and A.K. Misra
Software Cost estimation plays an important role in its development. A software cost (effort) estimation error can ruin any software. Number of models already exist that defines relationship between software size (input) and effort (output); however still these models fails to estimate the cost of software because of uncertainties, and imprecision associated with it. In this paper the parameters of existing cost estimation model (COCOMO) are tuned using improved Time Varying Acceleration coefficient based PSO and is tested on 10 NASA projects. The experimental results prove that the parameter tuned cost estimation model (COCOMO) has better estimation capabilities as compared to other existing cost estimations models.
Keywords: Particle swarm optimization, time-variant acceleration coefficient, software cost estimation, PM-person month, KLOC-kilo lines of codes, parameter tuning