Need a Data Scientist, Try Building a ‘DataScienceStein’
Organizations are finding that hiring qualified Data Scientist is a real challenge. Experienced Data Scientists are expensive and are usually employed elsewhere. This high demand, low supply economics is leading to a situation of the ‘haves’ versus the ‘have-nots’, where the larger, financially rich organizations in the ‘sexy’ industries are most capable of attracting and hiring data scientists, while the lesser companies will have to make do without one. Recent studies have indicated that there will be a significant shortage of data scientists for years to come.
Consequently, organizations need to look at new approaches to finding data scientists. Some are able to attract them with more than money like autonomy and professional development opportunities. Others are training current staff to become more data literate through professional development programs. Once trained, these individuals typically must work 12 to 24 months at the organization or have to “pay back” the amount spent on their training.
There is another approach that should be considered. It involves building your data scientist out of a team of people currently on staff or readily available in the marketplace. This is called the DataScienceStein approach modeled after Mary Shelley’s Frankenstein monster built from several human parts. In this case, building a DataScienceStein from a team with a variety of skills.
Some of the keys skills most required from a data scientist include:
- Data integration
- Advanced analytics
- Data visualization
- Industry or subject matter expertise
- Communication skills
- Programming skills
Other skills may be needed depending on the industry, company and the maturity or the organization’s analytics function.
To compensate for these skillsets a DataScienceStein team would need to include a:
- Data analyst who is responsible for gathering the data in an easily accessible location for use by other members of the team;
- Business analyst who understands the business and the data most relevant to the business leaders;
- Modeler responsible for creating and executing statistical models;
- Programmer to write the scripts in the preferred language to prepare, integrate and analyze the data for the modeler and the visualization specialist to use and the industry expert to review;
- Visualization specialist to translate the data results into visually engaging charts and diagrams;
- Subject matter or industry expert to provide insights on the data, industry and perspective on the results of models.
Many of these skills, like data analyst, business analyst and visualization specialist are being developed in certificate programs at schools like the University of Chicago’s Postgraduate Program in Data Science and Machine Learning (PGPDM) offered in partnership with Jigsaw Academy. The program is a blend of data science, business analytics, visualization and project management education with the application of advanced analytics models for artificial intelligence, deep learning and cognitive computing.
Larger teams and/or larger projects could also require additional programmers to write the code to access the data, quantitative analysts to help write scripts to access the data and project managers to keep the everything on track.
Two key skillsets that are extremely important to the effectiveness of the team are the subject matter or industry expert and the visualization specialist. Depending on the industry or business, having a person with extensive experience will help the team deliver more valuable analysis. This person provides the reasonableness factor by determining if the results of the analysis ‘makes sense’. The importance of these experts cannot be underestimated.
The visualization specialist is responsible for turning the data findings into graphics that non-technical people can easily understand and tells a story about the insights the data generated.
Using the DataScienceStein approach has several advantages to hiring the traditional data scientist, including:
- It broadens the applicant pool. By not having to find all of the skills in one person, but by searching for specific skills from a number of people, the organization has a better chance of identifying and hiring people to fill the positions. For example, data science is a fairly new function, so, as previously mentioned, it will take a while for the supply to catch up with the demand. Modelers on the other hand have been around for decades. The chances of finding a qualified modeler are many times better than finding a data scientist.
- Cross-functional learning. Members will naturally pick up skills from each other which will make them more effective at the work they are individually tasked to perform. This cross-functional learning will not only benefit the DataScienceStein team, but also the individual. By learning from others on the team, they are also gaining valuable skills to help them in their careers.
- Key knowledge is disseminated. Years ago, when data was hard to access and the tools were harder to use, there was typically one person with the skills to perform the analysis for an organization. If that person was injured or left the organization, all that knowledge would go with him/her. Now that the data is more accessible and the tools easier to use, several people can access the data and perform the analysis. A cross-functional team is even better situated to share this knowledge with each other and with those outside of the team, creating the opportunity to accumulate the information for others to easily access and utilize.
The main obstacle to the success of a DataScienceStein team or any data science project is the lack of leadership. Without clear direction and authority, the projects may end up in the vast nebula of projects launched by the organization that end up incomplete or ineffective because no executive leadership was provided. Leaders must have a vision for the team, be engaged with them and support them as needed.
After you’ve decided to take this approach the only decision left would be if you should model your DataScienceStein after the original “Frankenstein” character played by Boris Karloff who brings terror to the villagers or a Peter Boyle DataScienceStein from Mel Brooks’ “Young Frankenstein”, who can sing and dance for the audience. For the sake of your organization, you might want to choose the latter.
This article was authored by J. Bryan Bennett, who is the founder and Executive Director for the Healthcare Center of Excellence, a privately-funded research, training and consulting organization established to help healthcare organizations transition to be more analytics-focused. He is the author of the books, Prescribing Leadership in Healthcare: Curing the Challenges Facing Today’s Healthcare Leaders and Competing on Healthcare Analytics: The Foundational Approach to Population Health Analytics. Mr. Bennett is a highly requested national speaker on the strategic implementation and use of analytics in various healthcare areas, population health management and healthcare leadership.
He is also an adjunct instructor at the University of Chicago’s Graham School of Professional Studies, where he teaches Business Management and Project Planning in the Data Analytics for Business Professionals certificate international and domestic programs.
Mr. Bennett has a Master of Business Administration from Northwestern University’s Kellogg School of Management and a Bachelor degree from Butler University. He is a Certified Lean Six Sigma Green Belt, Certified Data Scientist, Certified Public Accountant and Certified Adjunct Faculty Educator.