By 2020, the number of data science and analytics job listings is expected to reach 2.72 million, according to a joint study by Business-Higher Education Forum, Burning Glass Technologies and IBM. What exactly do data scientists do and why are they being highly sought after? Gideon Mann (pictured above), Bloomberg's Global Head of Data Science, sheds light on those questions, and shared how his company is using data science.
CIO Asia: Since the data scientist role is still fairly new, can you share with me what a data scientist's typical day looks like?
Gideon Mann: The data scientist typically has three main functions:
1. Data analysis and modelling
2. Programming and implementation
3. Thinking about business goals
A data scientist needs to be thinking about what kind of business goals he/she is trying to achieve. What is the data that I need to collect? How do I take one of the models and push it out?
Bloomberg is a bit unusual in the data science space in that we have many strong programmers. Our core product is a software product. Unlike a pharmaceutical company or a retailer, a big part of what we deliver to clients is software. The data scientists we have are therefore mainly involved in delivering software. We can call them quantitative machine learning people, as opposed to business analysts.
What are the common misconceptions about data science/scientists?
The usual way companies think about the data science function is how to build this group in isolation, but that doesn't quite work. Data scientists need to be in the middle of product development. People don't always appreciate how important programming is to data science. In fact, many academic departments produce data scientists who are predominantly statisticians, and they can't be that effective.
What are the essential skills a data scientist should have in order to be successful?
In addition to having a sense of business and being able to communicate these priorities, a data scientist must be able to do modelling and programming. The other essential skill they should have is a curiosity to learn new things and be able to go outside his/her comfort zone to collect data. Things are changing so rapidly, so you can't afford to just stick with what you know.
Specific to your company, what is Bloomberg using data science for?
One of the areas of data science that Bloomberg has invested in is natural language processing (NLP), which is machine learning methods applied to text. We use NLP in many ways - to improve our operational efficiency around text processing, document structuring and extracting information from documents and news.
We are also using data science and machine learning to improve the usability of our infrastructure and software platform, transitioning from mnemonics to more natural language search interfaces.
Finally, we are using data science to develop new products. For example, we have some leading-edge news and social media sentiment products that enable clients to get a holistic and real-time view of chatter and sentiment analysis around a particular company or industry sector.
Sign up for Computerworld eNewsletters.