Using parametric regression and KNN algorithm with missing handling for software effort prediction

Estimating the software development costs, budget and resources such as the time and effort is one of the most important activities in the software project management. The error rate, at the estimating costs, has a sizable portion in success or fail of a project. In general, it is used from similar project histories for project estimation. One of the challenges in this approach is missing values. in this research, first, for handling missing values the K nearest neighbor (KNN) algorithm and Mean Imputation has been used, then for effort prediction, the parametric model based methods, the nonlinear and polynomial regression(quadratic) is used. The proposed method is performed on the CM1 dataset and the results show that the combination of KNN and nonlinear regression (quadratic) has the best response, signifying accuracy improvement and relative error reduction, in comparing with other approaches.