Subscribe to DSC Newsletter

Avengers vs. Superman – What should a Data Science team be made of?

how-to-form-a-world-class-data-science-team

The shortage of data scientists is a hindrance to the widespread adoption of analytics across many industries.

At the same time, the preponderance of multiple tools, techniques and knowledge base along with the rapidly changing data science landscape make it difficult for analytics leaders to hire the right talent for their data science team. The attrition rate in this industry is also high due to which employers end up spending a lot of time and money to hire a data scientist who may eventually not last for a long time in the organization.

In this blog post, I will show the best approach you can take to build a data science team.  Here is an analogy of Superman and Avengers, which might act as a good model for building a Data Science team.

If we notice Superman, what exactly makes him powerful? He has all kinds of things built within him. He has superhuman strength, X-ray vision, infrared vision, heat vision, telescopic vision, superhuman breadth, and hearing. Apart from this, he is quite fast as he has super-human speed; he stores and works on solar energy. Therefore, he is a one man- army. Now, who exactly are Avengers? Avengers are not as powerful as Superman. Each member of the Avengers team has one or maximum two powers like Superman. Some team members are not even superhuman. However, when we combine all the strength and abilities of each member of the Avengers team, this together can defeat the Thanos.

Now, let me correlate this to the Data Science team. A few days back, a renowned company approached me to help them recruit a data scientist. When I asked them the required skill sets, they mentioned the below list:-

  1. Ph.D. in Statistics/Economics.
  2. Proficiency in SAS, R and Python/Pandas.
  3. Well versed in Java, PMPL, Hadoop and Unix.
  4. Knowledge of Hive, Pig, Mahout and Spark.
  5. 5 years’ experience in healthcare domain and web analytics.
  6. Expert in Machine Learning environment.
  7. Knowledge of visualization tools.
  8. Project/Product Manager.

In short, they wanted a Superman with n number of super-powers. Would a Super man with these powers be a right fit? Alternatively, would a team of Avengers having one of the above qualities fit in? For instance, the avenger’s team would have one expert in statistics, one expert in Python and an expert in Java, so on and so forth. There is no open-ended strategy, that says hiring a Superman would be correct, or hiring an Avengers would be appropriate.

Let us look at the pros and cons of each.

Benefits of hiring a Superman:-

  1. He is a one-man army and can perform all roles independently, hence no need to hire another candidate.
  2. Since he is the only person, there is no threat of the silo mentality, i.e. the inability to share information across various departments.
  3. As he is not working in a silo, there is no need of co-ordination or hiring a Project Manager.
  4. As no team is involved, there is no need for team building activities.
  5. Possibly cheaper in the long run but when compared with a group of Avengers it is expensive as a single person to be hired.

Disadvantages of hiring a Superman:-

  1. As he has n number of qualities, it takes more time to find the right candidate.
  2. Retaining such candidates is also a challenge, as he may not stay long unless you have something good to offer him.
  3. “Kryptonite” could kill him.
  4. He might have some weakness.
  5. He may not scale easily as he has to handle several tasks at the same time.
  6. You have to trust his work, as you have no other option.
  7. He would be a Jack-Of-All-Trades but Master of none. He can provide you the breadth but not the depth of expertise required.

Let us look at the advantages of hiring using the Avengers model:-

  1. They are easier to recruit.
  2. As there could be n number of people having that specific expertise, it would also be easier to retain or replace them.
  3. Since they are easy to retain, this model would be scalable as you may have 5-6 people working on a single project rather than just a Superman.
  4. As they are masters in their field, they provide the depth required and the breadth comes along with the multiple experts in the team.
  5. They can provide faster turn-around time as multiple members are working on the same project.

The Cons of the Avengers model:-

  1. The silo mentality issue pops in, where each avenger would have a different perspective of looking and solving the problem.
  2. To avoid the silo issue, a good project manager is required to set common goals for the team.
  3. There would be too many moving parts, as too many people need to be kept updated about the project, hence coordinating all of them is essential.
  4. We need to analyze whether the sum of all parts is greater than the whole or is the whole greater than the sum of parts. This can only be realized after hiring the entire team.

After analyzing the pros and cons of both the models, how do we actually create the data scientist team? How do we understand whom to hire, when to hire and how to hire? To solve this question, we have created a pseudo-framework, where we have several questions, which a company needs to answer before initiating the recruitment process.

The Pseudo Framework,

DEFINE

  1. Why are you creating a data science team?
  2. What problems are you looking to solve?
  3. What goals do you plan to achieve by solving those problems?
  4. What kind of data do you have?
  5. The resources you have for creating a team

UNDERSTAND

  1. What is your organizations’ culture like?
  2. Does your company already have a data-driven culture?
  3. How involved is the management in using data for decisions?
  4. How skilled are people in your company at analyzing data?
  5. How is your organization structured regarding teams/department?

Moving forward on deciding which would be a good fit for your organization,

Superman is a good fit for following scenarios:-

  1. If your organization is small and, you don’t have the budget to hire an entire team to hire at one go.
  2. At the same time, your goals of solving your business problem and analyzing your data are not very clear, and you need this person to find those goals as well.
  3. As the Superman model is not scalable, you must have few problems to be solved and these problems must be interesting.
  4. You must have something special to offer this person to retain him.

Read more here

Views: 534

Comment

You need to be a member of Data Science Central to add comments!

Join Data Science Central

Follow Us

Videos

  • Add Videos
  • View All

Resources

© 2018   Data Science Central™   Powered by

Badges  |  Report an Issue  |  Privacy Policy  |  Terms of Service