Three Phases Young Decision and Data Scientists Go Through
Sometime around 2018, at an Opex Analytics company meeting, Jung-Hee Oh gave a short talk about the phases a decision or data scientist goes through. I thought it was a brilliant talk. It has stuck with me, and I wanted to get it written.
The ideas are all hers— and this is my recollection of it.
Phase One (in school), “The right answer.”
When you are in school learning about all these cool models and techniques, your focus is on getting the correct answer. You have a problem to solve. You want to pick the right model, correctly manipulate the data, find the answer, and earn the “A.”
This isn’t bad. You need to learn the material and build a foundation.
Phase Two (early jobs), “Work with cool technology.”
Often, the next phase is to find a job where you work with the latest and coolest technology. (When Jung-Hee gave this talk, that would have been deep learning; now, it is agentic AI and LLMs.)
Wanting to work with the latest and greatest technology isn't wrong. However, there are two things to watch out for.
First, it may bias your choice of where to look for work. You may rule out companies or organizations that look boring from a technology and tool point of view but would be a good fit for you.
Second, you may become unnecessarily frustrated and unhappy with your early jobs. If you aren’t working with cool technology, you may feel like you are falling behind or not doing what you want. This could cause you to miss out on good learning experiences or to switch jobs hastily.
Phase Three (the realization), “Work on things that add value.”
At some point, you realize that working on projects that add value is a lot more satisfying.
Working on cool technology that doesn’t lead to anything is ultimately not satisfying. On the other hand, there is a lot of satisfaction in seeing people use the things you create to make their lives easier and to make your organization better.
Thank you for the posting. At age 71, I have been involved in what we now call analytics for over 50 years, the core pattern hasn't change. Although I would add additional details - another time. One critical growth element in this path is diversification of your technical knowledge. If you are enamored with deep learning - take a look at MILP optimization or tools like Hexaly. On the stat side take a look at non parametric statistics. Second revisit the fundamental underpinnings of the analytical tools you use? For example in inventory why is the normal distribution often used?
This diversity puts you in the best position for step 3 - make a significant impact on organizational performance. Gary Sullivan, IBM retired, one of the best agents of change ever, observed you know your application has been successful, when folks can not remember life without the application. Great examples of step 3 can be found each year in the INFORMS Edelman competition. The 2025 competition is in Early April in Indianapolis. There are some great applications. This year, I am partial to Flipkart, since I am their coach. Flipkart is a great example of the importance of analytics without borders and understanding organizational dynamics to overhaul decision support.
Again thank you for sharing.