Clustering methodologies

Smekens, K. E. L., De Feber, M. A. P. C. and Seebregts, A. J. (2002). Learning in clusters methodological issues and lock-out effects. International Energy Workshop. EMF/IIASA, 18-20 June 2002, Stanford University. 

Clustering methodologies represent unsupervised analyses that are not appropriate for the incorporation of prior knowledge about the observations. Clustering is the process of grouping the data into classes or clusters so that objects within a cluster have high similarity in comparison to one another but are very dissimilar to objects... 

The clustering methodology is based on calculating the degree of similarity using PCA and distance similarity factors. Many researchers have used PCA with clustering to reduce the dimensionality of the feature space. The number of linearly dependent features is reduced and their scores are calculated. The scores are then used as new uncorrelated features that are clustered (61,62). 

First principle quantum chemical methods have reached the stage where they can now begin to provide reliable information on the structure, spectral and energetic properties of adsorbates on surfaces. Both the cluster, as well as the periodic band methods, were found to be quite successful. In contrast to early MO-based cluster studies, the DFT cluster methodology provides more accurate predictions, provided that special care is taken to optimize the structural configurations, the geometries, and spin states for all clusters. 

Another sophisticated investigation into the nature of monocyclic Cjo, using coupled cluster methodology, was presented by Watts and Bartlett [263]. Structure 9 was found to be a local minimum, while the fully symmetrical structure 8 shows one imaginary frequency and... 

Dubes, R. Jain, A.K. Clustering Methodologies in Exploratory Data Analysis . Advances in Computers 1980,19, 113-227. 

---