Performance Evaluation of the Data Clustering Techniques and Cluster Validity Indices for Efficient Toolpath Development for Incremental Sheet Forming
Abstract The goal of this research is to compare the data clustering techniques and cluster validity indices for feature-based tool path development, in case of incremental sheet forming process. The work compares the four most popular clustering techniques, i.e., partition-based (K-means), density-based (DBSCAN), variants of hierarchical clustering and graph-based (Spectral) clustering technique. Besides, for the quality assessment of the clustering solutions and to pinpoint the superlative validity indices, techniques like Calinski-Harabasz, Ball-Hall, Davies-Bouldin, Dunn, Det Ratio, Silhouette, Trace WiB, and Log Det Ratio are compared. The Single Linkage Hierarchical Clustering is preferred over the other variants as it detects the arbitrarily shaped clusters. After comparing it with DBSCAN, K-means, and Spectral clustering, it is found that DBSCAN is the best suitable technique for the proposed application. From the comparison of the internal validity indices, the following four out of eight techniques, Ball-Hall, Dunn, Det Ratio, Log Det Ratio indices are selected as they support the application. The outcome of this research would help in building algorithms for feature-based toolpath development strategies for manufacturing industry using data science and machine learning techniques.