Effects of Data Transformation on Cluster Analysis of Archaeometric Data

Petrographic and geochemical investigations of the late prehistoric ceramics from Sos Höyük, Erzurum (Eastern Anatolia)

“…There are different ways for data transformation to be applied in archaeometric investigations (see e.g., Bieber et al, 1976;Baxter, 1994Baxter, , 1995Baxter, , 2003bBaxter, , 2004Baxter et al, 2008;Baxter and Buck, 2000;Baxter and Freestone, 2006;Aitchison, 1986;Aitchison et al, 2002;Mucha et al, 2008).…”

The Destructive/Non-Destructive Identification of Enameled Pottery, Glass Artifacts and Associated Pigments—A Brief Overview

“…The development of X-Ray Fluorescence (XRF) instruments in the sixties, with the bore glass bead technique (boric acid or lithium tetraborate is added as a flux to solubilise about 1 g of ceramic in a Pt crucible and form a glass bead permitting quantitative composition measurement) [120][121][122][123] can be considered as the true onset of archeometry literature with the help of data processing (Principal Component Analysis, Cluster variation methods, etc.) [124][125][126][127]. The increasing sensitivity and versatility of instruments (Inductively Coupled Plasma-Mass Spectroscopy (ICP-MS), Energy Dispersive Spectroscopy coupled with a Scanning Electron Microscope (EDS-SEM), etc.)…”

Peer Review #1 of "Data-driven classification of the certainty of scholarly assertions (v0.1)"

“…Anything other than the 'Poor' category is considered in other studies to represent an acceptable level of agreement. (Deery et al, 2000;Lix et al, 2008) Clustering: As an initial step, due to the compositional structure of the data, it was necessary to perform a transformation of the data prior to Principal Component Analysis (PCA) and correlation statistical analysis (Mucha, Bartel & Dolata, 2008). Compositional data are data in which the sum of all components represents the complete set or a constant value (Mateu-Figueras et al 2003).…”

“…To estimate this, Hierarchical Clustering analysis (HCA) and Spearman correlation test were performed to determine certainty category association between questionnaires ( Fig. 3), using the shared classified statements in that category as the metric (Narayanan et al, 2011;Campbell et al, 2010;Sauvageot et al, 2013;Narayanan et al, 2014); though these constitute conceptually distinct analyses, we represent them in the same chart because the outputs are mutually supportive. Hierarchical clustering analysis (HCA) finds clusters of similar elements, while Spearman correlation coefficient considers the weight and direction of the relationship between two variables.…”