Building an organization to win with data

0

What kind of skills should organizations build to improve their data analytics capabilities? As organizations in Asia and elsewhere turn to data insights and create strong data cultures to gain an edge over their competitors, training has become an increasingly relevant topic.

And as we noted last year, citizen scientists working on dozens of projects can achieve cumulative victory on a massive front, allowing organizations to win big on data by starting small. But what skills should organizations train or hire their employees in: programming, business analysis, or statistics?

A question of practical knowledge

For at least one practicing data scientist with a Bachelor of Science in Reddita good background in statistics outweighs computer science, or else it believed. Until he took a master’s degree in statistics and his view of the world was completely “turned upside down”.

“Much of what we learn is completely useless for private sector data science, in my experience. So much useless math for the sake of math. Incredibly tedious calculations. Complicated proofs of irrelevant theorems… what’s the point? »

“There is hardly any work with data. How to train in statistics without working with real data? There is no real value to any of this. My skills as a data scientist [and] applied statisticians do not improve,” the poster wrote.

The post attracted nearly 200 comments, many of whom added thoughtful comments and shared their own experiences. Defending the role of statistics, some respondents argued that training in statistics provides intellectual enrichment and laid a deep foundation that helped them understand the “essentials” of machine learning.

Ultimately, courses that emphasize practical knowledge over theory are probably the most desirable for citizen data scientists. Seen in this light, then perhaps an introductory workshop that leverages organizational data could have a far greater impact than sending them off to week-long classes on deeply theoretical topics.

People skills matter

Effective communication is also vital. In addition to data analysis, findings and insights need to be communicated clearly and smoothly across the organization to have impact. Doing this well requires getting both technical and non-technical audiences to understand the implications. Crucially, they must then inspire disparate employees to take collective action.

In the field, this means identifying and empowering employees with strong interpersonal skills and data execution expertise. The former can also mean being able to say “no” to ideas that just don’t make sense or are doomed to failure.

The reproach of one poster on another Reddit thread summed it 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. ice cream with good accuracy because inputs are basically not very predictive of outputs.

Find success with data

For organizations to succeed with data, there is also no hesitation in developing organization-wide skills with data.

the Harvard Business Review (HBR) Noted that large companies ensure that as many stakeholders as possible have the data-centric skills and resources they need, instead of confining this expertise to the realm of specialists.

What kind of skills should organizations build to improve their data analytics capabilities? As organizations in Asia and elsewhere turn to data insights and create strong data cultures to gain an edge over their competitors, training has become an increasingly relevant topic.

And as we noted last year, citizen scientists working on dozens of projects can achieve cumulative victory on a massive front, allowing organizations to win big on data by starting small. But what skills should organizations train or hire their employees in: programming, business analysis, or statistics?

A question of practical knowledge

For at least one practicing data scientist with a Bachelor of Science in Reddita good background in statistics outweighs computer science, or else it believed. Until he took a master’s degree in statistics and his view of the world was completely “turned upside down”.

“Much of what we learn is completely useless for private sector data science, in my experience. So much useless math for the sake of math. Incredibly tedious calculations. Complicated proofs of irrelevant theorems… what’s the point? »

“There is hardly any work with data. How to train in statistics without working with real data? There is no real value to any of this. My skills as a data scientist [and] applied statisticians do not improve,” the poster wrote.

The post attracted nearly 200 comments, many of whom added thoughtful comments and shared their own experiences. Defending the role of statistics, some respondents argued that training in statistics provides intellectual enrichment and laid a deep foundation that helped them understand the “essentials” of machine learning.

Ultimately, courses that emphasize practical knowledge over theory are probably the most desirable for citizen data scientists. Seen in this light, then perhaps an introductory workshop that leverages organizational data could have a far greater impact than sending them off to week-long classes on deeply theoretical topics.

People skills matter

Effective communication is also vital. In addition to data analysis, findings and insights need to be communicated clearly and smoothly across the organization to have impact. Doing this well requires getting both technical and non-technical audiences to understand the implications. Crucially, they must then inspire disparate employees to take collective action.

In the field, this means identifying and empowering employees with strong interpersonal skills and data execution expertise. The former can also mean being able to say “no” to ideas that just don’t make sense or are doomed to failure.

The reproach of one poster on another Reddit thread summed it 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. ice cream with good accuracy because inputs are basically not very predictive of outputs.

Find success with data

For organizations to be successful with data, there is no escaping the need to develop organization-wide skills with data.

the Harvard Business Review (HBR) Noted that large companies ensure that as many stakeholders as possible have the data-centric skills and resources they need, instead of confining this expertise to the realm of specialists.

“[The] leaders view the use of data and analytics as deeply embedded in their operations, rather than keeping it siloed and limited to a few employees,” he said.

This means making data accessible not only to citizen data scientists and business leaders, but also to frontline staff. They also acquire data from customers and suppliers, with almost nine in 10 (89%) sharing their data in return.

As part of their efforts to democratize data, leaders are also twice as likely to enable remote access to data and store “a significant fraction” of their data in the cloud, the report noted. HBR report.

Here is. In addition to equipping workers with practical and relevant skills to manage data and appointing the right leaders to drive the organization’s data initiatives forward, data democratization and a cloud-centric approach to data are essential foundations. to be successful with data.

The rewards are worth it. According to the report, top performers in machine learning can achieve more than double the impact in half the time compared to the average business. And one suspects that this gap will only widen, not narrow, over time.

Paul Mah is the editor of DSAITrends. A former system administrator, programmer and professor of computer science, he enjoys writing code and prose. You can reach him at [email protected].​

Photo credit: iStockphoto/Christian Horz

Share.

About Author

Comments are closed.