Signal

* get up at 8 to 9am

* have breakfast if got up soon

* commute to office for 45 minutes, arrive at 10am

* work

* leave at 6.30 to 7pm

* arrive home 45 minutes later

* dinner

* random web browsing

* sleep at 12am

> wouldn't it be useful if I could just grab my User with ID 23456, and see a tree display of all the data referenced by it, rather than trying to piece this together by looking at various tables? What about its evolution in time? Wouldn't it be useful to just have a slider so that I could watch the data in this tree at previous points in time, or to be able to step through changes that were applied to it?

This! Looking at the data as a tree of references, or the data itself being hierarchical, is usually the most intuitive way I visualise it in my head. Same for the time, looking at the visualisation changing with time should be the way to visualize data evolution. In my work we deal with a lot of navigation logs, and what I miss the most is the ability to eaily visualize the navigation of a user and all of its attributes in the time, or how he steps from page to page. Or to visualize how the page attributes change with time, or the whole tree of url levels and how it evolves. The most challenging thing about it IMHO is how to correcltly handle not only one instance in the visualization but many, say millions, without forcing the user to pre-aggregate and loose all the interesring details.

cs231n by Andrej Karpathy : http://cs231n.github.io

Strange that Apache Flink and Google Dataflow don't figure in the Stream Processing list