Research on the prediction model of Neolithic archaeological sites based on GIS and XGBoost algorithm

This paper focuses on Shandong Province as the study area, collecting 1916 Neolithic archaeological sites (excluding burials) as the study sample. It randomly extracts 1916 non-archaeological sites as negative samples at a 1:1 ratio. Eight geographic environmental influencing factors, such as elevation, slope, aspect, profile curvature, planar curvature, micro-geomorphology, slope position, and distance to water boundaries, are selected to construct the index system for the archaeological site prediction model. The GIS spatial analysis method is employed to build the predictive model of archaeological sites using the XGBoost algorithm. The model predicts the spatial extent of potential archaeological sites and conducts an important factor analysis. The results indicate that: (1) the archaeological site prediction model constructed by the XGBoost algorithm achieves high accuracy, with an AUC value of 0.85 in this study; (2) the model trained with optimal samples categorizes the results into low, medium, and high grades, and concludes that the archaeological sites are mainly distributed in the plains area; (3) according to the XGBoost algorithm to rank the importance of the influencing factors, slope position, microgeomorphology, and elevation are the top three factors influencing the siting of Neolithic archaeological sites in the study area. It is found that the XGBoost algorithm has good stability and prediction ability, and the constructed model provides a new research method for the prediction of archaeological sites, and provides an important technical support for field archaeological excavation, which reveals the relationship between the site selection of ancient human beings in the Neolithic era and the geographic environment.