For anyone that’s spent a couple of weeks in front of a computer trying to make sense of some jumbled up data, you’ll have come to the same conclusion as the rest of us: at times, patterns in noise can be hard to characterise.

You’ll find that if a certain variable in a puzzle moves up or down, then the results change considerably — and at times, you’ll even struggle to try make sense of this yourself. Now I’m a visual thinker so I need to be able to see my results, which at times is near impossible when you have a large number of variables.

If you want to support a business case: there’s no better way than using visuals.

However, this problem isn’t new and a lot of smart people have written about the importance of characterising data the recent Coronavirus outbreak has even further accentuated the importance of effective display (reference this).

So in what comes, I offer advice on how to display data at different levels and pay particular attention to displaying joint dynamics: so how specifically how the variables are related. This is different to just plotting information from two variables — it’s trying to characterise more from a probabilistic sense.

Think about a business case where you want to invest into a new marketing strategy as you believe it’ll result in an increase in revenue. The easiest way to win that business case is by proving that investing into your business case will directly result in an increase in revenue.

I mean you can show t-stats and complicated statistical experiments but let’s be honest: people prefer charts to numbers.

Ultimately, you want to prove the dependence between your marketing strategy and business revenue and you want this ‘dependence’ to be robust. Well if it’s robust, shouldn’t it be easy to see?

#machine-learning #artificial-intelligence #software-development #data-science #python