The more you think about what it takes to make better decisions, the more you realise it can require a huge range of disciplines from psychology to data science to user experience. Having worked in data science on anomaly detection, business analysis on root cause analysis and enterprise data architecture on enabling information to be a better asset; much of that experience can be summarised as helping clients make better decisions.
Bernard Marr concludes that there are three ways a business can get value from data: use it to make better decisions, use it to improve operational efficiency, or monetise the data itself by selling it or the insights it creates. My interest has always been with the first.
After 17 races across the USA and Canada, involving over 30 drivers in cars going over 200mph, the NTT Indycar championship concluded in September. During the race weekend, each team had to make decisions about which tyres to use, how much fuel to start the race with, how many pit stops were needed, how much downforce from the wings would be required for that particular track, what improvements to the car were needed after qualifying.
The significance of a wrong decision, or even a late decision, can determine the race outcome. A good example was happening on the other side of the world; I was at the spectacular Formula 1 Grand Prix in Singapore on Sunday, September 22, where Ferrari earned an unexpected first and second place on the podium due to pitting earlier than Mercedes.
To support such decision making, there are different technological implementations that can help. At the Laguna Seca race track at the weekend, the Chip Ganassi team supported by NTT DATA were relying on their Race Strategists, Mike Hull and Barry Wanser, to decide when to pit their drivers. A huge amount of data is made available to them to support this decision but there are many different ways to exploit it.
One technological option is a ‘black-box’ machine learning model which learns from previous races on how quickly tyres degrade based on track temperature, driver style, circuit layout etc. and suggests to the race strategist (perhaps through a natural language bot) which laps the drivers should pit on.
The model would need to be complex enough to adapt to real time event data during the race (such as when other teams are pitting). In this case, the decision making would be quick but opaque.
On the other hand, a simple screen showing a table of all driver times and tyre choices could help race strategists make the decisions themselves based on their experience and instinct. However, this relies on the ‘humans in-the-loop’ to compute a vast amount of data which may lead to errors or simply take too long.
Ultimately, the right solution requires consideration of the user experience for the precise way that Mike Hull and Barry Wanser make decisions. This allows appropriate data visualisations to be developed which cater for the amount of detail that they would like to see. We all make decisions in different ways.
I was struck when reading about the various factors that can affect them like being hungry or leaning to your left hand side!). We all suffer from biases. Although not fully understood, we all use a blend of gut feelings and emotions as well as logic and reason when making decisions, and this varies from individual to individual.
Therefore, we need to consider psychology too when implementing technology. For team Chip Ganassi, we need to examine end user perspectives in addition to their biases so that we augment together the best of human intelligence with the best of artificial intelligence.
It is important therefore not to consider data science as an isolated discipline even if it requires some very specialist skills. Rather, data scientists should collaboratively support the business’s overall goal with whatever decision needs to be made.
To do this, experts in customer experience are needed to create intuitive designs, data visualisation experts are needed to create actionable intelligence from data, whilst business consultants and analysts are needed to get to where the real problem and value lies through quantitative analysis and ethnographic studies.
Making better decisions is a complex task and requires a much broader perspective than we often give it. But many of our clients are finding that it is worth it whether or not a world championship is on the line.
To find out more, visit uk.nttdata.com/Impact
References
Cooper, B. B. 8 Things You Don’t Know Are Affecting Our Choices Every Day: The Science of Decision Making. Retrieved from https://open.buffer.com/decision-making/ Marr, B. (2017).
Data Strategy: How to Profit from a World of Big Data, Analytics and the Internet of Things Bernard Marr.