What separates a great data scientist from a good one?

Good Data Scientist: Python, R, SQL, engineering skills, Spark + longer list of technical skills

Great Data Scientist: Subset of the above + Empathy, Tailored communication, Active Listening, Adaptability

The skills to level yourself up from a good data scientist to a great one can be learned with practice, and I describe four ways that can help. For each of the four examples, my work with undergraduates helped me enormously.

Your stories and experiences may be different from mine, but you may recognize common themes. Perhaps the model and metrics looked great to you but the stakeholder was unconvinced; or there was organizational pushback to putting your model in production; or there were engineering constraints you had not taken into account and you felt like you wasted work or there was a lot of rework, etc.

A key thing to keep in mind before moving forward — similar to technical skills, building up your people skills is an investment in yourself and does take time and effort. This is worth it!

1. Show, not tell

There is nothing more demoralizing for a professor than a student who regularly sleeps in your class which starts at 11:30 a.m. This happened to me and I was determined to keep this person awake and hopefully teach them something. Determined and prepared, I ceremoniously filled a rusty, metal bucket with water, tied the bucket handle to a rope and then spun the bucket over my head to demonstrate the centripetal force. I had the rapt attention of over 60 Gen-Xers all waiting for their professor to be doused by a bucketful of water. Once everyone was properly awake, it was a lot easier to introduce involved calculus.

The skill to capture and maintain interest from a stake holder has been key in my Data Science career. This skill has helped me tailor my language, gauge my audience and demonstrate concrete examples of use cases without losing my audience in the technical jargon of a model.

Work your way through the The Craft of Scientific Presentations: Critical Steps to Succeed and Critical Errors to Avoid by Michael Alley. This is a completely practical guide and has some great pointers on tailoring to an audience.

Screenshot of the Table of Contents from the The Craft of Scientific Presentations by Michael Alley — https://www.amazon.com/Craft-Scientific-Presentations-Critical-Succeed/dp/0387955550

2. Managing expectations and creating transparency

Setting up murky grading expectations generally invites the wrath of a class-full of 19 year olds and does not lead to good learning outcomes. I learned early on as an undergraduate instructor the importance of setting expectations early and often. It was much easier to communicate information at a regular cadence — three times a week, every week, fully knowing that deadlines could move if a health situation came up or there were other extenuating circumstances.

The practiced ability to set expectations and create transparency with my manager and my colleagues meant that I was able to contribute to the team right away, and when I made a commitment, I was generally able to follow through unless there were clear extenuating circumstances. Communicating at a regular cadence also meant that my colleagues were quickly able to point out when something was not right, which led to early diagnosis and quicker resolution of blockers. I found that this had an empowering effect on the team. It was rare that I was the only person facing a particular technical issue and during my “over-communication” people would chime in with solutions or indicate solidarity by demonstrating that they were having the same problem.

Keep a daily log of what you have accomplished for the day, similar to a daily standup with yourself. Share it with your manager and colleagues. When annual performance review time rolls around refer back to your log to steer yourself and your manager away from recency bias. Over time, the list will also help you be reflective and help find blindspots, get better estimates for work, and make sure projects are aligned with your career goals.

3. Leverage organizational support

I developed and taught a hands-on, lab based class that taught physics to people that did not care about physics and had no mathematical background but were required (re:forced) to take a lab based course as a graduation requirement. A legally blind student walked into my office and told me that she had heard good things about my version of the course and would like to take it. I know she felt my incredulity even though she could not see me. I reminder her that she would be required to build projects and work with circuits to get a passing grade. She was persistent. Not knowing what to do, I walked straight into my chairs office and told her that I would like to help a blind student take my course but I had no clue how. My chair was incredibly supportive with words and with a budget. She connected me with disabilities services and a talented TA; and found the budget to hire the TA. A legally blind young woman thrived in my hands-on, project based class.

I learned the skill of being persistent and asking for organizational support for asks that I felt were impossible for me to handle on my own. I found that my managers and leads are very open to clearing roadblocks if I concretely communicate what I need — even if the ask seems impossible at first.

If you have a creative idea that can deliver business value but don’t have the resources you need — speak up! Talk to your colleagues in engineering, product, and data science. They may have a broader view of resource allocation, may have more access and be completely supportive.

4. Empathy, Safety and Active Listening

I would often get students that wanted to work on improving their grades, when this happened I was able to leverage a solution that was right for them. Sometimes that solution was a metacognition exercise where we would go through a quantitative form to measure their effort versus result to locate areas where they were not preparing enough. My goal was to witnessed a person have an “Aha” moment that diagnosed their issue and then they had something concrete to overcome. Other times the solution to a performance issue was to ask them to get a good nights sleep and find ways to reduce testing anxiety. These different solutions require some focussed listening, empathy and creating an environment where a person feels safe sharing their struggles. For example, poor listening leads to asking an over-prepared student to study more leading to more anxiety and worse test grades.

In the corporate context, the skills of empathy, safety and listening translated over seamlessly. Data scientists also appreciate a judgement free zone where they could fail and learn to get back up again. I am often able diagnose issues for colleagues mostly by listening and providing a judgement free zone. I enable individuals to arrive to a realization on their own. The temporary discomfort of sitting with the issue in a consequence-free setting is enormously helpful when coming up an effective, and lasting solution.

You can start your journey with the help of this 2015 PyCon talk by Sasha Laundy — Your Brain’s API: Giving and Getting Technical Help. Sasha primarily focusses on how to ask for help without being scared and how to provide help without being a jerk.

Thank you to Michele Callson, Pauline Ugalde, Kyle Takeuchi, Grace Chesmore and so many students that were all my mentors. They gave me the courage to change career paths. Thank you to John Birmingham, Keith Griffioen, and Betty Young for showing me that empathy is my strength and not my weakness.

Data Science @WalmartLabs. Physicist/DataScientist/ Wife/Mom/ChurnedAcademic. Find out more about me here! https://www.linkedin.com/in/suchetajawalkar/