Tennessee Leeuwenburg https://2016.pycon-au.org/schedule/78/view_talk Most 'data science' projects fall into just a few well-understood design patterns. This talk de-mystifies what those patterns are, how to use them practically, and how to get to grips with your data. We'll a look at how to understand the input/output structure of the models, how to design a reasonable 'experiment', and how to get started. We'll look at getting to grips with problems by simple data sets that can fit entirely on-screen, designing the basic 'form' of the machine before levelling up to bigger data and badder algorithms.
All of this will be shown using Python tools, libraries and running code.