3 Ps for Chief Data Scientists
Promoting data science, picking realistic projects, and people skill development are the top three priorities for chief data scientists right now.
Companies are starting to make bigger bets on artificial intelligence (AI) and machine learning (ML), and organizations are expecting more from their chief data scientists. A large part of the role of a chief data scientist is managing senior management's expanded expectations about what artificial intelligence and machine learning can do -- and cannot do -- to move an organization forward.
Now more than ever, the chief data scientist has to be an evangelist for the technology. The best chief data scientists work within the organization to help business units apply artificial intelligence and machine learning to their projects.
The key priorities for chief data scientists right now are promoting data science, picking the right projects, and mentoring people to run those projects.
#1: Promote Data Science: Be an Evangelist and a Realist
In addition to being an evangelist, a chief data scientist needs to be a realist. Roughly 60 percent of all data science projects end in failure. The role of the chief data scientist is to help pick the projects that are most likely to help the organization with the lowest risk of failure and decide how to allocate resources to those projects.
In addition to working with senior leadership to talk about what's possible for the organization, a good chief data scientist needs to spend time with the business units understanding their needs for the next few years. By working directly with business units, the chief data scientist can expand AI and ML capabilities throughout the enterprise.
Picking the right projects to pursue will be critical because resources are limited.
#2: Pick Your Projects: Make Sure They're the Right Projects
A chief data scientist needs to determine which new projects are integral to the company's growth, which will improve efficiencies, and which are merely nice to have.
For mission-critical projects that can push the organization forward, the chief data scientist needs to build out an internal team to tackle them. Here, a chief data scientist wants to move activities from the experimental stage to fund real projects that can add to a company's bottom line.
For projects that improve efficiency, the chief data scientist should work with their teams to find a way to buy the best technology available. There are wonderful tools to help you with images, text, time-series analysis, etc., but to understand how to use those tools, the chief data scientist needs to stay current with what tools are coming on the market to evaluate best-in-class.
For projects that are nice to have, the chief data scientist should determine whether the enterprise has the resources to push those projects forward or whether it's worth it to bench those projects because the risk associated with advancing a non-essential project could outweigh any possible gains.
#3: Prioritize People Skills: Good Mentors Are Hard to Find
Picking the right projects to pursue will be critical because resources are limited. There simply aren't enough good data scientists to perform all the work the organization wants to complete. Smart chief data scientists will use the next six months to set up training programs for staff at internal business units. They will need to hold training classes with people throughout the enterprise to help them gain the skills needed to deliver projects at the business level.
It is typically hard for leaders who do not have data science experience to know if data science can help with a business problem they want solved, and the risk and cost associated with solving it with a data science-based solution. As we are emerging into a post-pandemic world, chief data scientists need to help senior leaders understand how artificial intelligence can help with long-term business strategies while keeping them informed that many of these projects come with uncertainty, and the desired outcome may not come to pass.
Final Thoughts
Data science projects are not like normal software development because they are driven by data and therefore fraught with uncertainty. When you are starting a data science project, you don't know if you can achieve the results you want to achieve. The chief data scientist must have a realistic idea of what the risks are and be able to communicate those risks throughout the enterprise.
About the Author
Ira Cohen is co-founder and chief data scientist of Anodot, in charge of inventing and developing its real-time multivariate anomaly detection algorithms. He holds a Ph.D. in machine learning from the University of Illinois at Urbana-Champaign and has over 15 years of industry experience. You can reach the author via email, Twitter, or LinkedIn.