AutoML Time Series Forecasting for Infrastructure Resources

Explore a 34-minute conference talk on applying AutoML time series forecasting powered by Spark to plan IT infrastructure resources. Discover a framework that focuses on forecasting time series metrics for computation and storage resources like CPU, memory, and disk sizes. Learn about the innovative use of Spark-based AutoML and big data distributed computation to automate and linearly scale up the fitting and forecasting of millions of time series models in parallel. Understand how this forecasting framework addresses long-term planning for maintaining adequate supply equilibrium and guides daily dynamic provisioning of computing and storage resources. Examine two empirical applications where thousands of time series models are automatically built and used to forecast and plan short- and long-term resource demand. Presented by Heping Liu, Senior Principal Engineer at Workday Inc., this talk offers valuable insights for IT professionals and data scientists working on infrastructure resource planning and optimization.