Anomaly Detection for Time Series Data
Predicting traffic is another use case of anomaly detection (i.e., consumer traffic, network traffic) . Usually traffic is normal but there are instances where traffic can spike or drop unexpectedly. In this case study, we will look at a dataset that studies traffic for taxis in New York City (NYC). The time period spans from June 30th 2014 to Jan 31st 2015 (i.e., 6 months). During this time there were 5 events where the amount of traffic changed significantly(NYC marathon, Christmas, New Years Day, Snowstorm, and Thanksgiving).