Distortion function and clustering for local linear models

The vector quantization principal component analysis (VQPCA) was analyzed for better understanding of distortion function that was used for clustering the data space. VQPCA involved a two-step procedure, namely a clustering of the data space into a several partitions such as, projection and Euclidean partition, and the application of PCA in each local partition. VQPCA was applied for the reconstruction of dynamical responses and was more effective tool than conventional PCA. The projection partition offered a better reconstruction of the dynamics of the portal frame than the Euclidean partition.