Welcome to the Machine: New Techniques in Predictive Modeling for Improving Data Quality in Zooarchaeology

Summary

This is an abstract from the "Recent Advances in Zooarchaeological Methods" session, at the 88th annual meeting of the Society for American Archaeology.

Taxonomic identification is a key goal of faunal analysis, but few controls are in place to ensure data quality. Comparative collections and identification guides offer valuable information; however, the validity of faunal identification can be questioned without assessing each feature’s utility for differentiating taxa. Analysis of biometric data allows zooarchaeologists to evaluate criteria from identification guides. This study evaluates well-known criteria for distinguishing tarsals of deer (genus Odocoileus) from pronghorn (Antilocapra americana) published by Barbara Lawrence in 1951. We measured tarsals from reference collections and built models using techniques in supervised machine learning to assess these criteria. Assessing and improving the use of quality-control methods with robust predictive modeling workflows is a new way to approach data quality. The result is a set of identification criteria that are rigorously verified and designed to handle new data from diverse contexts, which can significantly improve data quality in zooarchaeology.

This Resource is Part of the Following Collections

• Recent Advances in Zooarchaeological Methods   •
• Society for American Archaeology 88th Annual Meeting, Portland, OR (2023)

Keywords

Spatial Coverage

Record Identifiers

Abstract Id(s): 36727.0