Subscribe

Data science is a team effort

Mervyn Mooi
By Mervyn Mooi, Director of Knowledge Integration Dynamics (KID) and represents the ICT services arm of the Thesele Group.
Johannesburg, 27 May 2016

Data science is pitched as the sexiest job of the 21st century and businesses around the world are on a quest to harness the data skills that promise to boost their revenue. But businesses need to be aware that data science, like most technology disciplines that came before it, is over-hyped and often misunderstood.

While it's true that data science is becoming the 'brain' that informs business of the future, it is not entirely new.

Data science is not a single discipline, and no one data scientist can be expected to deliver on its promise alone. While it's true that data science is becoming the 'brain' that informs business of the future, it is not entirely new - it is the evolution of traditional BI, analytics, database architecture and more.

Data science in a mid- to large-sized enterprise must comprise up to 20 different, specialised disciplines working together to take data from source to delivery, in such a way that it effectively informs business strategy.

These disciplines include:

  • Establishing requirements - the business analyst must play a role in determining why business needs what data.
  • The 'sanity' division - the information analyst, data steward or systems analyst must carry out the data discovery, confirming the necessary data is available and can be extracted without risk or compromising governance or SLAs.
  • The extraction 'how-to' team - the systems analyst, programmers and database administrators work within corporate policies to extract the data, and must massage, franchise or prepare that data. Preparing this data for use also involves data warehousing, data integration specialists, data analysts and architecture teams designing the tools and processes.
  • The analysis experts - only after the data has been prepared can analysts step in to apply analytical models. At this stage, the involvement of business strategists and risk analysts is important.
  • Project managers and data science heads will also be needed, to oversee data science processes from source to delivery.

With so many specialists required to manage vast volumes of data and extract business value from it, the task is clearly beyond the capabilities of a single data scientist. And South African companies are starting to realise this.

Instead of seeking the 'magic bullet' of data science, they are starting to build multi-skilled data science teams to collaborate closely and extract the real value of data. As these data science teams grow and proliferate, we can expect to see the lines between business and data science becoming increasingly blurred, until data science evolves into something more accurately called business science.

In future, no business will be able to compete without an effective data science or business science team. The time to lay the groundwork for such a team is now, and there is no better place to start assembling the data science team than within the business's existing skills base.

By upskilling in-house BI professionals, database administrators, systems architects and others to better understand analytics and business strategy, the forward-thinking business will position itself to harness the power of information in future.

Share