Visualized Performance Evaluation of 3D CAD Software Based on Hierarchical Clustering Considering the Uncertainties in Insufficient Test Data

The performance of 3D CAD software is usually evaluated based on the test data obtained through automatic test. The uncertainties in the test data caused by the fluctuation of computer performance and testing tools are unavoidable in repeated tests, and it is probable that the test data are insufficient due to limited repetitive tests. To provide intuitive and convincing performance evaluation results based on the insufficient test data, this paper proposes a visualized performance evaluation method for 3D CAD software based on hierarchical clustering considering the credibility of performance indexes. Firstly, the performance indexes are described as interval numbers, and their credible scores are calculated based on the interval numbers. Subsequently, the credible scores are integrated into a series of credible score vectors. With the similarity between every two vectors measured by Euclidean distance, the credible score vectors are merged sequentially according to the principle of minimum increment of overall variance. Finally, a heat map describing the hierarchical clustering process is generated, which provides an intuitive and credible performance evaluation results of the 3D CAD Software.