Preamble

One thing I’ve learnt the hard way is the importance of communicating data science results. As scientists and engineers, we focus on the methods, the engineering and the solutions. There’s where most of us have a passion for. Communication is hardly a passion, there’s no engineering craft in there - or is there?

One of the teams that placed ahead of us in the CIKM-DHL Hackathon did basically dashboard data visualisation the whole way, whereas we invested 90% of our time into developing predictive algorithms. Sure, you can argue the benefits of one over the other. Visualisations have the primary purpose of communication, whereas the primary purpose of algorithms is to deliver advanced models.

There’s no point in comparing which is more meaningful as priorities change depending on the operational context. But the bottom line is when decision-makers are the key stakeholders (or paying your salary) - a simple algorithm and advanced visualisation more often than not trumps an advanced algorithm and simple visualisation. What matters? Actionablility matters - “what is one recommendation that you can make to me that I can do tomorrow?”

Engineer’s Perspective

Of cos the guys-in-suits don't get it. Tch. They dont get anything.

Stakeholder’s Perspective

And you tell me this for?

Conclusion

Like it or not, we need to be able to communicate Data Science results for survival. There’s no vessel for your algorithm otherwise. If you won’t touch the stuff, find a UI person to partner with.