Big data is at least four different problems (2015) [video]
youtube.com
youtube.com
It's not to say they don't exist, obviously analytics is super important, but after being involved in it for 10+ years, it is rare that I have seen any game changing results come from big data analytics teams. Lots of small incremental improvements, sure, but I wonder about the ROI given the infra costs.
Or analytics for sports? The sabermetric revolution uses big data for video analysis now - something like 6 TB per game.
Flexport might be another example. Although I’m not privy exactly to how they use data.
- Large amounts of source data
- Large amount of processing/analytics of data (ML and other models)
- Large flow of data (data per time unit)
- Diverse input sources (combining data from different sources and formats)
Obviously these aren't mutually exclusive.