Companies Need to Develop a Data Science Strategy

By Samuel Greengard  |  Posted 2017-07-26 Email Print this article Print
 
 
 
 
 
 
 
Data science strategy

Once business leaders understand how data science can make a difference, an enterprise can formulate a plan for putting data science methodologies to work.

A multidimensional approach may span disciplines such as economics, anthropology, statistics, analytics, information technology and artificial intelligence (AI). The goal is to gain a broader and more holistic view of data—and then put it to work in new and improved ways.

A key, says Accenture's Kanioura, is to start with a business problem and let project teams define the direction of the initiative and the IT framework needed to support it. She believes that it's important to view data science in a flexible way, explore different scenarios, and use modeling techniques to understand how applying data in different ways could lead to different outcomes and results.

What's more, the entire process must have validation methods built in. "Someone in the business unit, or multiple business units, must check to make sure that the organization has defined the best solution for the business problem," Kanioura says.

Understanding the Art in the Science

It's important to recognize that data science is essentially a human endeavor that requires specific skills and knowledge. Data scientists command hefty salaries—a 2017 Glassdoor survey found that the median base salary for professionals in the field is $110,000, and experienced data scientists earn much more. Retaining these specialists can also prove challenging. Finally, there are also limitations to what data science can deliver.

As a result, Kanioura says, it is crucial to establish boundaries for how an organization will use data science and how project teams will explore topics. "You have to give them the freedom to explore, but you also have to give them direction and focus," she points out.

Not surprisingly, some organizations turn to outside providers to address data science needs—particularly as today's business problems become more complex and incorporate diverse areas such as the internet of things (IoT) and AI. Yet, regardless of whether an enterprise builds a data science team internally or turns to outside service providers, TCS' Shroff says it's important to recognize a few key things.

First, data science doesn't come in a specific shape or size, and there's no template for how to focus an initiative. Second, data science requires creativity and intuition; it ultimately falls somewhere between art and science. Finally, success is as much about combining the right skills and people as it is about specific data.

Nevertheless, data science will play an increasingly important role in transforming organizations in the years ahead.

"It is very important to avoid the media hype about data science and AI," advises Shroff. "There are very real opportunities to use data to drive decision making, automation and insights. As digitization sweeps across industries and sectors, understanding business context is critical to success. Data science plays an important role in generating value."



<12
 
 
 
 
Samuel Greengard writes about business and technology for Baseline, CIO Insight and other publications. His most recent book is The Internet of Things (MIT Press, 2015).
 
 
 
 
 



















 
 
 
 
 
 

Submit a Comment

Loading Comments...
Manage your Newsletters: Login   Register My Newsletters