A New Empirical Model to Increase the Accuracy of Software Cost Estimation

We can say a software project is successful when it is delivered on time, within the budget and maintaining the required quality. However, nowadays software cost estimation is a critical issue for the advance software industry. As the modern software’s behaves dynamically so estimation of the effort and cost is significantly difficult. Since last 30 years, more than 20 models are already developed to estimate the effort and cost for the betterment of software industry. Nevertheless, these algorithms cannot satisfy the modern software industry due to the dynamic behavior of the software for all kind of environments. On this study, an empirical interpolation model is developed to estimate the effort of the software projects. This model compares with the COCOMO based equations and predicts its result analyzing individually taking different cost factors. The equation consists one independent variable (KLOC) and two constants a, b which are chosen empirically taking different NASA projects historical data and the results viewed in this model are compared with COCOMO model with different values of scale factor. In this paper the author analyze more than 250 projects collected from PROMISE repository. The effort variance is very low and the proposed model has the lowest Mean Magnitude of Relative Error (MMRE) and RMSSE.

Keywords:

Lines of code , Software cost estimation , Magnitude of Relative Error , Mean Magnitude of Relative Error , Pharma Research and Early Development