Predictive Modelling of Settlement Patterns in the Córdoba Province (Argentina, South America).

IZETA , ANDRES; BILOTTI GIACOMO; CATTANEO ROXANA

Oakland

Congreso; 51st International Conference CAA (Computer applications and Quantitative Methods in Archaeology).; 2024

CAA

Predictive modelling is an open-box that includesall type of statistical analyses that can be usedto “predict” an outcome. Although in very simpleforms these models have always been used inarchaeology, their formalisation dates back tothe processual revolution. Today, understandinghuman-environmental relationship holds acentral role in academic research, especially inenvironmental sciences, and archaeology hasa central role in it. However, bridging the gapbetween often-inaccessible data and globalresearch interests remains a challenge.In recent years, researchers at the UniversidadNacional de Córdoba and the CONICETcompiled an extensive, standardised and upto-date database of the archaeological sitesthroughout the Province of Córdoba (Argentina).The database contains 2390 sites, classifiedby chronology and typology. This is the firsttime that in this region a freely accessible andwell-maintained database is available, pavingthe way for many types of spatial analysesand modelling. Furthermore, the availability ofgeomorphological and environmental variablesand satellite imagery increase the possibilities ofmore accurate modelling.Our presentation aims to present the results ofthe first ever-carried predictive modelling in thisregion. Specifically, we focus on site locationpatterns and site-environmental relationship inorder to understand past population’s behavioursin terms of settlement choice and its variationthrough time. Additionally, a more formalunderstanding of these dynamics will allow moreprecise modelling of archaeological potential,with positive outcomes in terms of heritagemanagement and selection of suitable areas forsurveys.We approach the predictive modelling byconsidering archaeological sites as pointprocesses and carrying out Point Pattern Analyses(PPA), studying the underlying first and secondordereffects generating the observed processes.We are interested in site intensity as a functionof spatially continuous covariates, such asenvironmental and geomorphological variables.These variables are then used as predictorsin the fitting of a Point Process Models to ourdata. Furthermore, our model incorporates theinteractions between sites, in order to explore therelationship, stability, and changes in settlementpatterns using a hybrid approach. We recognizethe inherent challenge in fully disentanglingfirst and second-order effects and addressingresearch biases. Therefore, we integrate thesecomponents in a more holistic and sustainablemodel.Notably, all analyses presented herein areconducted using free and open-source software,complemented by open data sources. Thisapproach not only enhances the transparency andreproducibility of our research but also extends its71educational impact to our respective institutions.By making these resources accessible, we aimto allow an increasing number of students andresearchers to engage in quantitative analyses inarchaeology in Argentina and South America.