Keeping Up With Data — Week 20 Reading List
Giro d’Italia is one of my favourite cycling races. Maybe because we used to take a couple of days off, pack the bikes and drive to the Italian Alps to try compare our climbing abilities with the pros on the same roads. We didn’t pay much attention to data. We barely planned the routes, ignored the distance and the vertical gain. But now data is everywhere. You can see many real-time stats of the riders at the Giro. Just like NHL coaches can track the data about their players. And while my work is addressing business problems with data analytics, the application of data science in sports (and cycling most of all) is a wonderful blend of my ‘loves’.
Three very different topics this week, but all are somewhat related to decision making. (But what isn’t, right?)
- Modern Recommender Systems: What to recommend to whom is a very common business problem. So, it’s no surprise that it’s fuelling the race to the best recommender system. It is interesting to read about the evolution from content-based and collaborative filtering to deep-learning recommenders. Firstly, because we can see how individual methods were focusing on the weaknesses of their predecessors. And secondly because it provides a nice intuition behind all these complex systems. (Maximilian Beckers @ Towards Data Science)
- The problem with AI developer tools for enterprises (and what IKEA has to do with it): AI developer tools or ML platforms are entering many companies. What’s hard about that? Firstly, there is no dominant design — every cloud provider and many start-ups are aiming to solve similar problems with vastly different solutions, which confuses a lot of people. Secondly, all these tools have different user and developer experience, which often comes with a love-or-hate relationship. And thirdly, there is a strong ‘IKEA effect’ when the internal teams want to (co-)create their tools. For these reasons, the choice of the ML platform often leads to endless discussions and when a decision is finally made a new decision maker, with different preferences, enters the stage and you go back to the drawing board. So, let’s hope the decisions around the ML platforms will become easier, faster, and less driven by personal taste. (Clemens Mewald @ Towards Data Science)
- Dashboard Psychology: Effective Feedback in Data Design: What is the psychology behind “what you measure, you improve”? How do numbers and charts compel people to act? And what psychology of feedback has to do with that? The article shows how a graph going up can make people keep going or make important changes just based on the colours; how we can use comparisons and contrasts and benchmarks to make the message stronger (and influence the action); or when to use positive and negative feedback. Effective dashboards provide effective feedback. They make the target audience more committed and focused. (Eli Holder @ Nightingale)
That’s it for the reading. Let me now (Thursday evening) jump on a bike and squeeze in a quick ride before dinner. And yes, I’ll be tracking the data!