The first challenge is deciding on a window, and what size window we need to accommodate to make sense of the incoming data. By window, we mean storing the last n data points. For streaming data, it is rare to actually process records one at a time. One of the most common ways to process streaming data is to process them in a window. This can either bundle data points into a group and process them as a unit or it can be a changing group of n data points (or x time) where older records are discarded and newer records added.