The Joy of Stats

We’ve had a stats love-in this week, with a gathering to review Hans Rosling’s recent BBC4 program The Joy of Stats and a lunchtime debate about data, statistics and the role they play in understanding our world.

Rosling shows that by making your data sing using basic statistical analysis and visualisations you can pull out interesting facts about the world, your neighbourhood or your customers. At Seren we are interested in how the world works from a Consumer’s perspective and also how Customers react when complicated information is provided to them. In our roundtable discussion this week the team highlighted the importance of not only finding a killer stat but also using both qual and quant data to understand ‘why’ you might be seeing a particular trend.

As well as basic stats Rosling talked about how mathematicians use statistical distributions to model behaviour. I’ll bet that the Normal distribution will represent height distribution in your office (it does in ours but with a very large standard deviation). When I first watched the Joy of Stats programme just before Christmas I laughed out loud when seeing someone talk passionately about my own favourite, the Poisson distribution, which is used in queuing theory. In my early days in statistics I used this distribution to predict traffic and pedestrian flow in Glasgow.

By understanding and modelling how the world works you can begin to predict what future events will take place and plan for them. It was great to see Rosling talk about complex data in a way that all statisticians should do; using simple language and simple analysis in a focused, relevant but maybe above all passionate way. Have you ever seen anyone get so existed about finding out how Sweden’s GDP developed in the last 200 years?

Rosling has found a beautiful and insightful way of visualising his passion; global healthcare statistics and how the world has developed in the last 200 years. Since he debuted his talk at TED in 2006 he’s became one of the movers in the information visualisation/infographics craze. The main thrust of the infographics movement isn’t actually to draw pretty pictures but to utilise the explosion and availability of data that is collected about our world and derive insight more easily. In Rosling’s talk there were examples of how San Franciscans can have immediate access to crime statistics in their neighbourhood which they use to ensure the police services are accountable (and to avoid a few areas too no doubt). As another example have a look at Tim Berners-Lee talking about how open, shared & visualised data improved the ability of aid organisations to respond to the Haiti earthquake in 2010. This mimics another one of Rosling’s examples but from way back in 1858. Florence Nightingale is famous in statistical circles not as the Lady of the Lamp but for her famous Rose charts. A simple statistical visualisation that did so much to improve the healthcare levels in hospitals across Europe.

Back to those Poisson distributions. Several years ago as a student in Glasgow I did a project with the help of the City Council to simulate traffic flow in the city, developing a bit of software with a few flashing icons to help me statistically optimise traffic light signals outside the university. I used some survey data and a Poisson distribution to simulate the frequency of cars and people arriving at traffic signals.

Our model was un-erringly accurate at predicting when people and cars would arrive at the junction and could predict the optimal time lights should stay green and what type of crossing pedestrians may need. The Council used a similar method across the whole of the city with the aim of improving signals and making traffic safer for pedestrians. At the time Glaswegians had the highest rates of pedestrian accidents in the UK. ‘Why’ was that? Could we get the context behind the stats? No quantitative data that was publicly available that explained this trend but it was an open secret that it was due to Glaswegians consumption of certain liquids, which affected their ability to get home safely.

Stats are lovely and beautiful but make sure you’re always be able to explain your findings!

This article was written by John D’Arcy.

Leave a comment

*