Wavelet transforms based ARIMA-XGBoost hybrid method for layer actions response time prediction of cloud GIS services

Abstract Layer actions response time is a critical indicator of cloud geographical information services (cloud GIS Services), which great significance to resource allocation and schedule optimization. However, since are highly dynamic, uncertain, uncontrollable, the layer influenced by spatiotemporal intensity concurrent access intensity, posing significant challenges in predicting action time.To predict more accurately, we analyzed data association services. Furthermore, based on characteristics long-term stable trends short-term random fluctuations series, wavelet transforms-based ARIMA-XGBoost hybrid method for proposed improve one-step multi-step prediction results time.We generate multivariate series feature matrix using historical value time, predicted linear component, non-linear component. There no need meet traditional assumption that nonlinear components additive, minimizes model’s requirements enhances its flexibility. The experimental demonstrate superiority our approach over previous models