Imputing incomplete software metric database using iterative nearest neighbour based algorithm

Abstract

Data sets are playing an important role not only for the construction of effort precision model but also in building software cost estimation models. Knowing the estimated cost at the early project would constitute so much to the software development cycle. Therefore, with the existence of missing values in the data sets, it will be thorny to achieve all these. The purpose of this paper is to review and implement the iterative nearest neighbour based data imputation technique for handling incomplete software metric database. The experiments are carried out using real database for the input of the algorithm for the purpose of relating the missingness data in the real database to examine the accuracy of the imputed data compared to the actual data.