It is important to segment time series into groups and model each group separately to enhance forecasting accuracy overall. But what does this look like in practice? Why not use an automated machine learning (AutoML) package or an auto-forecasting tool and let it do the hard work for you? We know that the best modeling technique is always based on the data we are dealing with. Machine learning methods have been proven useful for forecasting purposes, but they are computationally intensive and many times fail to outperform in accuracy simple statistical methods. This session explored how demand classification in SAS® Visual Forecasting can group time series based on their historical demand patterns, allowing users to apply the most appropriate modeling technique to each segment. This optimizes the forecasting process, increases accuracy and speed, and leads to better decisions in a cost-effective way in the cloud.
Presentation slides are attached to this post.
