Dirty data science: Machine learning on non-curated data

The author (Gael Varoquaux) is deeply involved in numpy/Sklearn and a talented researcher. Quite an impressive guy worth following.

The slides were pretty good. I just wish they were shared in a different format and/or via a different medium.

Trying to make a dataset on cost overruns/project management at work currently which has required quite a bit of manual work looking up project information in documents not exposed to the database.

I've looked through the slide deck and hopefully this will help me figure out some improvements to the way my data is structured.

Is there a soundtrack/transcript to go with this?

+1 very interested to hear this talk

I googled the talk name and found these:

https://www.youtube.com/watch?v=dw5u4nth6_M

https://www.youtube.com/watch?v=BsDeG3jQ61s

Neither seems to be _exactly_ the same deck, but lot's of overlap - should be pretty close.