Not a surprise, but the story is a good small example of the problem, written out with enough detail for non-experts to follow. Answering a very simple question necessitated the use of a mix of skills like interviewing the homeless, persuading a government agency to release data and hacking the structure of Excel documents.
"getting the data, ingesting the data, using deep domain knowledge". My point is/was that the super duper ML and visualisation is secondary (initially). What I've seen is that the above list is table stakes, after which (quite a long time after) someone realises that they can do something interesting with analysis. The next step is that we realise that. everything. we. thought. about. this. is. wrong.. Then much interesting thing!
