"Mathematical analysis is becoming ever more useful when dealing with large amounts of archaeological data, due to the precision and certainty with which results can be produced. This presentation will propose the use of new mathematical... more
"Mathematical analysis is becoming ever more useful when dealing with large amounts of archaeological data, due to the precision and certainty with which results can be produced. This presentation will propose the use of new mathematical tools in deciphering and dealing with archaeological data, with a specific focus on the Naïve Bayes Classifier and the promotion of its wider use by budding archaeologists. The ‘Bayesian’ approach was first proposed in the early ‘90s, by the most well known of archaeological statisticians (Orton, 1992:139, Buck et. al., 1996:1), though at that time the lack of computational power available made use of the classifier prohibitively difficult. Today, a Naïve Bayes Classifier can be utilised by anyone with a computer, without any need for particularly specialized computer skills. Programs such as Orange use a graphical interface as a way to circumvent the need for specific mathematical knowledge of the process, and the use of this program will be detailed in the presentation. The Naïve Bayes Classifier is most useful in attempting to identify unseen patterns in a large amount of data, such as a spreadsheet or database with thousands of entries. The analysis of the ~2700 rune-stones in Sweden as accomplished in my Honours thesis will be used to illustrate the ease with which this tool can be utilised, as well as the many situations for which use of the Naïve Bayes Classifier is appropriate."
The great motorway research and construction investments have brought and are still bringing a huge set of new data. In 2019 alone, one million new archaeological artefacts were sourced. Therefore, there is a problem of systematic and... more
The great motorway research and construction investments have brought and are still bringing a huge set of new data. In 2019 alone, one million new archaeological artefacts were sourced. Therefore, there is a problem of systematic and comprehensive development of the obtained sources, in which statistics may be helpful. The article introduces selected statistical methods and shows examples of their use. It focuses on their usefulness in archaeological research, and thus it may become a boost for their wider use in the development of archaeological sources
According to recent literature, correspondence analysis is the method of choice for frequency seriation. However, this does not consider the effects of data heterogeneity or typology on the orderings produced by this method. This relates... more
According to recent literature, correspondence analysis is the method of choice for frequency seriation. However, this does not consider the effects of data heterogeneity or typology on the orderings produced by this method. This relates to a more fundamental issue of how to evaluate the effects of heterogeneity and typology on seriation results, as well as how to determine which of a set of seriation algorithms produces the more likely seriation ordering on a particular data set, and if so, why? In this paper we present a new methodological framework that: (1) identifies which parts of a data set are amenable to seriation, (2) identifies the likely number of minimally heterogeneous components in a data set in a way that is sensitive to emic distinctions, and (3) operationalizes a Bayesian stochastic seriation model to evaluate which of a set of seriation algorithms produces the more likely ordering on a given data set. When this new framework is applied to heterogeneous mortuary data set from a coastal Chumash village in southern California, the results suggest that correspondence analysis does not produce a likely seriation ordering.
