Classifying incomplete software engineering data using decision trees: An improved probabilistic approach

An attempt is made to address the problem of classifying incomplete software project data. The essence of the approach is the proposal that software effort can be predicted in probabilistic terms given incomplete software project data using decision trees. This approach is based on the a priori probability of each value determined from the instances at that node of the tree that have specified values. The proposed approach exploits the total probability and Bayes' theorems, and it has three versions. We evaluate our approach using data from a multidimensional software house from the point of view of their effects or tolerance of incomplete test data. Experimental results are reported, showing the effectiveness of the proposed approach, and also in comparison with Quinlan's fractioning of instances or cases (probabilistic) technique.