Tag
This paper proposes a two-stage integrated forecasting model using XGBoost to predict CPU workload in private clouds by first forecasting customer service requests, achieving high accuracy with SMAPE below 7% for most applications.