Eoin Hurrell, Data Scientist
Each time you buy something on Amazon.com or watch a movie on Netflix, you are adding a little more information about yourself so, the next time you log on, there are personal recommendations.
They might not always be accurate (since when did Twihards watch the Vampire Diaries?) but a vast amount of data analysis is carried out and computer algorithms are applied to calculate these ‘guesses’.
This is the intersection of data and human behaviour. It is a space where data scientists search for and analyse patterns to make sense of buying patterns. This information is used by companies to sell you what they think you’re more likely to purchase.
It’s not only used by internet companies; supermarket loyalty cards are an early example of recording consumer behaviour patterns.
What does a data scientist do?
Eoin Hurrell is a freelance data scientist who carries out consultancy work. “I specialise in recommendation and prediction, as seen on Netflix where movies are suggested for you.
“My education is in computer science, but what’s interesting about data science as a field is its variety. It combines practical implementation skills like coding with maths and statistics skills, as well as domain expertise, so there is quite a bit of variety”.
Eoin says that there are many different algorithms for prediction, which can be very accurate when well designed.
“It can take as little as 10 movie ratings to start offering personalised suggestions in a system like Netflix. The more data it gathers, the more accurate it will get at predicting your taste”.
Making better recommendations
What about the future of personalised suggestions? Known as “predictive analytics”, companies such as IBM use health records to prevent conditions like heart failure by catching clues to its development in existing patient data.
“The power of analytics is in finding patterns in large amounts of data, and we are getting better at that,” explains Eoin. “On the data side, we may be able to detect something (for example Facebook page ‘likes’ can be indicative of everything from political leanings to sexual preference), but the real insight is whether it would be useful, tactful or beneficial to do anything about the insight gained, which is the human side of the challenge”.
Eoin is right. A 2013 study analysed the Facebook activity of 58,000 volunteers and found some interesting results. For example, those who ‘liked’ thunderstorms, science and curly fries were slightly more likely to be above average intelligence!
Check out this IBM page on data science for more information.