Abstract: 

"How do you write 'good' code?
How do you write maintainable code?
How do you write code that your colleagues can understand, use and modify safely, without risk of it breaking?
Importantly, how can you do this in a professional environment where there is rarely the luxury of extra time for 'a big clean up'?

These questions are by no means trivial and the issues they touch can be just as important to a team of Data Scientists as the analysis and algorithms they are producing.
In fact it is a real surprise that these matters rarely get more than a short mention in the training of a Data Scientist.

Well, the community of software developers has been pondering these questions for a long time and have come up with various different answers.
'Test Driven Development' is one answer, formalised by Kent Beck in the 1990s.

In this live coding session Robert Hardy will code up a simple Bayesian classifier in Python, using a TDD approach.
That in itself will be interesting to people unfamiliar with the principles behind a Bayesian classifier.
However, Robert will really be showing how the TDD approach offers a framework that can improve our chances of delivering good code that we can be proud to share with our colleagues.

Bio: 

Robert Hardy has spent over 12 years working in the finance industry as a quant. He likes to build full-stack numerical applications -- from data pipelines to mathematics to UI -- and has done this mostly in Python and Elm recently.

Robert is the organizer of Full Stack Quants, a meeting of like minds held monthly at Code Node in central London.