Forecasting volunteer grid workload using Holt-Winters' method

Abstract

Resource management can be improved by study and analysis of workloads; which will further help to predict the future workloads. In volunteer grid resources and workload submitted are unpredictable which makes resource management more challenging. In the paper, an attempt has been made to analyze trend and seasonality of volunteer grid workload using Holt-Winters' method. The forecast of workload that can be submitted to volunteer grid has been performed by varying seasonality and observed its impact on forecast values. The results indicated that the workload forecast can be improved by varying the smoothing parameters (α, β and γ) depending on the original dataset available and variant of Holt-Winters' method used.