Analogy-Based Software Quality Prediction with Project Feature Weights

摘要

This paper presents analogy-based software quality estimation with project feature weights. The objective of this research is to predict the quality of project accurately and use the results in future predictions. The focus includes identifying parameters on which the quality of software depends. Estimation of rate of improvement of software quality chiefly depends on the development time. Assigning weights to these parameters to improve upon the results is also in the area of interest. In this paper two different similarity measures namely, Euclidian and Manhattan were the measures used for retrieving the matching cases from the knowledgebase to increases estimation accuracy & reliability. Expert judgment, weights and rating levels were used to assign weights and quality rating levels. The results show that assigning weights to software metrics increases the prediction performance considerably. In order to obtain the results, we have used indigenous tools.