The catastrophic consequences of not being awesome at plots
“Hey boss, we made a totally rad deep learning algorithm! It trawls through the internet and literally tells the future!”
“Yeah, cool. But do you have a plot I can show the board?”
Data visualisation is the capstone of data science. As businesses collate massive and disparate data streams, and as algorithms become more complex, communicating results has become more important and more challenging than even before. We need to to start placing as much importance on accessible visualisation as we do on database architecture or algorithm design.
In this talk I will present some core visualisation principles that I have developed over 15 years of experience visualising (literally) astronomical datasets, carbon emissions, behavioural analytics, even the odd basketball game. These can be employed by any data scientist to help their data tell the right story to the right people.
Outline/Structure of the Talk
The main message is that data science can be let down by inadequate data visualisation. Once I introduce this topic I will motivate it by offering some examples of poorly though out data visualisation. I will continue by establishing some principles that explain where these visuals went wrong. Along with this I will outline examples of my own work and explain my design decisions. Then I will bring all this together into a toolkit for analysts to employ in their day to day professional life.
Data practitioners will come out of this session with a toolkit for data visualisation. It will cover the why and when to use what sort of visual, when to annotate, when to employ interactivity, and more.
Data analysts and data craft managers.
Prerequisites for Attendees
This will be accessible to all data practitioners.