How would your approach change if this data were streaming in real-time rather than provided in a static CSV?
