In this paper

we extracted 11 evaluation factors based on DEM

geological map

remote sensing image

geotechnical investigation data of power grid

etc.

and built the quaternary spatial information prediction model based on eXtreme Gradient Boosting(XGBoost) algorithm. Then the model was applied to Xuanzhou District

Xuancheng City

Anhui Province

and the spatial distribution of the Quaternary soil thickness in this area has mapped. The results show that the prediction accuracy of the total samples is 86.81%～96.13%

the overall classification accuracy is 91.15%

and the Kappa coefficient is 0.88. The overall learning and prediction ability of the model are strong

the prediction accuracy of the model is high

and the prediction results have high credibility. The model has a certain theoretical basis and a good application prospect in the prediction and extraction of Quaternary spatial information.