Expert systems archeological predictive model

John Ripy, Ted Grossardt, Michael Shouse, Philip Mink, Keiron Bailey, Carl Shields

Research output: Contribution to journalArticle

3 Citations (Scopus)

Abstract

This paper reports on the deployment of a predictive model that combines spatial analysis and fuzzy logic modeling to translate expert archeological knowledge into predictive surfaces. Analytic predictive archeological models have great utility for state departments of transportation, and some states have invested millions of dollars in such models. However, classic statistical modeling approaches often require too much data and create questions about whether areas are categorized as low probability because (a) there are no sites or (b) no surveys have been conducted there. However, this process can build robust models around typically sparse archeological data and is not subject to spatial bias. These models are intended to lower overall project costs by identifying corridors with a lower probability of having archeological sites, not to supplant field surveys once a corridor has been chosen. Five influencing factors were defined by archeologists and were calculated with the ArcGIS platform. The archeologists then informed a fuzzy logic induction process that was mapped to output probability functions. These data were geocoded into ArcGIS output surfaces that showed the probability of encountering artifacts. The predictive results were tested through a blind control protocol against cleansed archeological data. These models were shown to perform as well as or better than traditional statistical models and required much less data. The Kentucky implementation includes the superior predictive coverage and, more important, a suite of tools to allow the ArcGIS-competent archeologist to design and execute new modeling routines or to build new models. The availability of higher-quality geographic information systems data will also allow archeologists to update the model.

Original languageEnglish (US)
Pages (from-to)37-44
Number of pages8
JournalTransportation Research Record
Issue number2403
DOIs
StatePublished - 2014

Fingerprint

Expert systems
Fuzzy logic
Geographic information systems
Availability
Network protocols
Costs

ASJC Scopus subject areas

  • Civil and Structural Engineering
  • Mechanical Engineering

Cite this

Ripy, J., Grossardt, T., Shouse, M., Mink, P., Bailey, K., & Shields, C. (2014). Expert systems archeological predictive model. Transportation Research Record, (2403), 37-44. https://doi.org/10.3141/2403-05

Expert systems archeological predictive model. / Ripy, John; Grossardt, Ted; Shouse, Michael; Mink, Philip; Bailey, Keiron; Shields, Carl.

In: Transportation Research Record, No. 2403, 2014, p. 37-44.

Research output: Contribution to journalArticle

Ripy, J, Grossardt, T, Shouse, M, Mink, P, Bailey, K & Shields, C 2014, 'Expert systems archeological predictive model', Transportation Research Record, no. 2403, pp. 37-44. https://doi.org/10.3141/2403-05
Ripy, John ; Grossardt, Ted ; Shouse, Michael ; Mink, Philip ; Bailey, Keiron ; Shields, Carl. / Expert systems archeological predictive model. In: Transportation Research Record. 2014 ; No. 2403. pp. 37-44.
@article{c74250450abd4d2e80026dba25696f9f,
title = "Expert systems archeological predictive model",
abstract = "This paper reports on the deployment of a predictive model that combines spatial analysis and fuzzy logic modeling to translate expert archeological knowledge into predictive surfaces. Analytic predictive archeological models have great utility for state departments of transportation, and some states have invested millions of dollars in such models. However, classic statistical modeling approaches often require too much data and create questions about whether areas are categorized as low probability because (a) there are no sites or (b) no surveys have been conducted there. However, this process can build robust models around typically sparse archeological data and is not subject to spatial bias. These models are intended to lower overall project costs by identifying corridors with a lower probability of having archeological sites, not to supplant field surveys once a corridor has been chosen. Five influencing factors were defined by archeologists and were calculated with the ArcGIS platform. The archeologists then informed a fuzzy logic induction process that was mapped to output probability functions. These data were geocoded into ArcGIS output surfaces that showed the probability of encountering artifacts. The predictive results were tested through a blind control protocol against cleansed archeological data. These models were shown to perform as well as or better than traditional statistical models and required much less data. The Kentucky implementation includes the superior predictive coverage and, more important, a suite of tools to allow the ArcGIS-competent archeologist to design and execute new modeling routines or to build new models. The availability of higher-quality geographic information systems data will also allow archeologists to update the model.",
author = "John Ripy and Ted Grossardt and Michael Shouse and Philip Mink and Keiron Bailey and Carl Shields",
year = "2014",
doi = "10.3141/2403-05",
language = "English (US)",
pages = "37--44",
journal = "Transportation Research Record",
issn = "0361-1981",
publisher = "US National Research Council",
number = "2403",

}

TY - JOUR

T1 - Expert systems archeological predictive model

AU - Ripy, John

AU - Grossardt, Ted

AU - Shouse, Michael

AU - Mink, Philip

AU - Bailey, Keiron

AU - Shields, Carl

PY - 2014

Y1 - 2014

N2 - This paper reports on the deployment of a predictive model that combines spatial analysis and fuzzy logic modeling to translate expert archeological knowledge into predictive surfaces. Analytic predictive archeological models have great utility for state departments of transportation, and some states have invested millions of dollars in such models. However, classic statistical modeling approaches often require too much data and create questions about whether areas are categorized as low probability because (a) there are no sites or (b) no surveys have been conducted there. However, this process can build robust models around typically sparse archeological data and is not subject to spatial bias. These models are intended to lower overall project costs by identifying corridors with a lower probability of having archeological sites, not to supplant field surveys once a corridor has been chosen. Five influencing factors were defined by archeologists and were calculated with the ArcGIS platform. The archeologists then informed a fuzzy logic induction process that was mapped to output probability functions. These data were geocoded into ArcGIS output surfaces that showed the probability of encountering artifacts. The predictive results were tested through a blind control protocol against cleansed archeological data. These models were shown to perform as well as or better than traditional statistical models and required much less data. The Kentucky implementation includes the superior predictive coverage and, more important, a suite of tools to allow the ArcGIS-competent archeologist to design and execute new modeling routines or to build new models. The availability of higher-quality geographic information systems data will also allow archeologists to update the model.

AB - This paper reports on the deployment of a predictive model that combines spatial analysis and fuzzy logic modeling to translate expert archeological knowledge into predictive surfaces. Analytic predictive archeological models have great utility for state departments of transportation, and some states have invested millions of dollars in such models. However, classic statistical modeling approaches often require too much data and create questions about whether areas are categorized as low probability because (a) there are no sites or (b) no surveys have been conducted there. However, this process can build robust models around typically sparse archeological data and is not subject to spatial bias. These models are intended to lower overall project costs by identifying corridors with a lower probability of having archeological sites, not to supplant field surveys once a corridor has been chosen. Five influencing factors were defined by archeologists and were calculated with the ArcGIS platform. The archeologists then informed a fuzzy logic induction process that was mapped to output probability functions. These data were geocoded into ArcGIS output surfaces that showed the probability of encountering artifacts. The predictive results were tested through a blind control protocol against cleansed archeological data. These models were shown to perform as well as or better than traditional statistical models and required much less data. The Kentucky implementation includes the superior predictive coverage and, more important, a suite of tools to allow the ArcGIS-competent archeologist to design and execute new modeling routines or to build new models. The availability of higher-quality geographic information systems data will also allow archeologists to update the model.

UR - http://www.scopus.com/inward/record.url?scp=84903760483&partnerID=8YFLogxK

UR - http://www.scopus.com/inward/citedby.url?scp=84903760483&partnerID=8YFLogxK

U2 - 10.3141/2403-05

DO - 10.3141/2403-05

M3 - Article

AN - SCOPUS:84903760483

SP - 37

EP - 44

JO - Transportation Research Record

JF - Transportation Research Record

SN - 0361-1981

IS - 2403

ER -