Building an Organization to Win With Data
What type of skills should organizations strengthen to enhance their data analysis capabilities? As organizations in Asia and elsewhere turn to data insights and build strong data cultures to gain an edge over their competitors, training has emerged as an increasingly pertinent topic.
And as we noted last year, citizen data scientists working on scores of projects can achieve a cumulative victory on a massive front, allowing organizations to win big at data by starting small. But which skills should organizations train their employees in or hire for: Programming, business analytics, or statistics?
A matter of practical knowledge
For at least one practicing data scientist with a Bachelor of Science on Reddit, proper training in statistics trumps computer science, or so he believed. Until he took up a Master of Science in Statistics and had his worldview completely “flipped” around.
“Much of what we're learning is completely useless for private sector data science, from my experience. So much pointless math for the sake of math. Incredibly tedious computations. Complicated proofs of irrelevant theorems… what's the point?”
“There's basically no working with data. How can you train in statistics without working with real data? There's no real-world value to any of this. My skills as a data scientist [and] applied statistician are not improving,” the poster wrote.
The post attracted close to 200 comments, with many adding thoughtful comments and sharing their own experiences. Defending the role of statistics, some respondents argued that training in statistics offers intellectual enrichment and laid a deep foundation that helped them understand the “nitty-gritty” of machine learning.
Ultimately, courses that favor practical knowledge over theory are probably the most desirable for citizen data scientists. Viewed that way, then perhaps an introductory workshop that leverages data from the organization might have a far bigger impact than sending them away for week-long courses on deeply theoretical topics.
People skills matter
Effective communication is vital, too. Apart from analyzing the data, the findings and insights must be clearly and fluently communicated across the organization to make an impact. Doing this well entails getting both technical and non-technical audiences to understand the implications. Crucially, they must then coax disparate employees into taking collective action.
On the ground, this means identifying and empowering employees with strong people skills and good data execution expertise. The former might also mean being able to say “no” to ideas that just don’t make sense, or which are destined to fail.
The gripe of a poster on another Reddit thread summed this up perfectly: “I don't care if you have 300 million data points recording people's eye color and favorite ice cream flavor, you still can't use eye color to predict ice cream flavor with any good accuracy because the inputs are fundamentally not very predictive of the outputs.”
Finding success with data
For organizations to succeed with data, there is no shying away from the need to build organization-wide competency with data.
The Harvard Business Review (HBR) noted that leading companies ensure that as many stakeholders as possible have the data-centric skills and resources they need, instead of keeping this expertise within the domain of specialists.
“[The] leaders view the use of data and analytics as deeply embedded to how they operate, rather than keeping it siloed and restricted to a few employees,” it said.
This means making data accessible to not just citizen data scientists and business leaders, but also the frontline staff. They also acquire data from customers and suppliers, with almost nine out of 10 (89%) sharing their data back.
As part of their data democratization efforts, the leaders are also twice as likely to enable remote access to data and store “a significant fraction” of their data in the cloud, noted the HBR report.
There you have it. Apart from equipping workers with relevant, practical skills to manage data and appointing the right leaders to push the organization’s data initiatives, data democratization and a cloud-centric approach to data are vital foundations to succeeding with data.
The rewards are worth it. According to the report, top performers in machine learning can have more than twice the impact in half the time compared to the average company. And one suspects that this gap will only grow larger, not smaller, over time.
Paul Mah is the editor of DSAITrends. A former system administrator, programmer, and IT lecturer, he enjoys writing both code and prose. You can reach him at [email protected].
Image credit: iStockphoto/Christian Horz
