Big Data with APT
On the 24th of October OXWIB hosted Applied Predictive Technologies (APT), a strategy consulting company based in Soho, London. ‘Big data consulting’ may not quite jump out at you (the more mathematically minded are exempt), but we were eased into what exactly it is that APT does, primarily through a series of mini-stories. They began on a historic note.
When the cholera epidemic raged in 1854, John Snow, a physician, plotted cases of cholera on a map to find that by far the most cases were concentrated around a pump on a certain street. A brewery not so far away had much fewer cases of cholera around it, but crucially also had a different water supply. Snow used this data to argue, correctly, that cholera was water-borne rather than air-borne.
This kind of ground breaking data is what drives APT. Analysing data, and creating the occasional ‘map’ or ‘hotspot zone’ like this is APT’s selling point. Another example closer to home was through Netflix, which had initially started out by selling and renting DVDs. Netflix has since boosted its growth by using algorithms to up its subscriptions, seen in the form of a ‘thumbs-up or thumbs-down’ system to attribute a ‘% match’ for its TV shows or movies. How did such ideas come about? In part through a 2007 lottery that granted a million dollars to anyone who could come up with an algorithm to increase subscription rates by 10%.
Similarly, data analysis consulting is what really defines APT. Their annual 24 hour hackathon challenge, for example, tried to assess what most affects access to clean drinking water in Liberia, with factors such as population growth, climate and temperature and population density playing crucial roles in what affected supply.
In the office in fact, APT does even more. They recognise that some of the limitations of big data is the difficulty of ascertaining causation versus correlation, and consequently of anticipating relevant behaviour despite knowing the existence of some kind of correlation. APT tries to resolve this. In the food industry, it helped approve the churning out of McDonald’s all-day breakfast, after concluding that it would be a viable, successful option. M&S’s cauliflower steaks, on the other hand, was deemed to be a potentially imprudent investment, and accordingly that was recommended to be phased out. Both these decisions were made after extensive experimentation and using a statistical ‘test and learn’ approach.
APT offers several internships for analysts. Those unaccustomed to the deep rigours of data analysis also have reason to cheer - their business consultant roles don’t require prior analytical data experience, with training been given on the job. It doesn’t take much data analysis to conclude that this might be a worthwhile career to consider.