Section: New Results
Day-ahead Time Series Forecasting: Application to Capacity Planning
Participants: C. Leverger, V. Lemaire, S. Malinowski, T. Guyet, L. Rozé
In the context of capacity planning, forecasting the evolution of server usage enables companies to better manage their computational resources. The work in  addresses this problem by collecting key indicator time series. The article proposes a method to forecast the evolution of server usage one day-ahead. The method assumes that data is structured by a daily seasonality, but also that there is typical evolution of indicators within a day. Then, it uses the combination of a clustering algorithm and Markov Models to produce day-ahead forecasts. Our experiments on real datasets show that the data satisfies our assumption and that, in the case study, our method outperforms classical approaches (AR, Holt-Winters).