This project illustrates how to work with sliding windows in Flink using a count-based slide and a time-based window length (e.g. W(l=30 min, s= 1)). Consider a stream of user event data where you need to process the events of the last 15 minutes whenever a new record arrives.
The provided example application reads Uber trip data that summarizes the Uber pickups during July 2014. It is structured as follows:
"Date/Time","Lat","Lon","Base"
For every ten records that arrive the stream processor, the count of pickups per base of the last thirty minutes is emitted.