How to Be a Kick-Ass Data Scientist: Steps 1-3

December 15, 2020

Bo Chipman, SVP & Client Partner


Data Science is hot. There is not enough talent in the market to fill demand, and new academic programs are producing data science graduates at a record pace. At BLEND360, we hire our fair share of these graduates. They are an incredibly talented bunch with loads of potential. It is our job to transform them from new graduates into Kick-Ass Data Scientists. This blog highlights some of the lessons we have learned in the process.

Steps 1-3

Everyone knows data management absorbs most of an analyst’s time. Instructors will tell you this step requires 70% to 80% of total effort. What they may not tell you is the success of your data science project starts before you touch the data. Understanding the business problem is foundational to all projects, and it is a step less seasoned data scientists often take for granted. All data scientists must have a diverse skill set. The best begin every project as Anthropologists, interviewing their clients to gain a deep understanding of their universe. They establish the framework for success by doing the following things well.

1. Ask Questions

Our culture is infatuated with solutions. In our rush to fix problems, we often fail to comprehend the nuances of the problem and the surrounding ecosystem. Data scientists are not immune to this phenomenon. It manifests itself as a tendency to debate methods early in the process, often before the problem is well understood. The all-to-frequent result is an elegant solution that never gets implemented.

It is easy to avoid this pitfall by asking questions – lots of question – of your client. Clients generally have sophisticated knowledge of business processes. They know what they need, even if they do not know how to build it, and they can generally tell you what will and will not work. Ask them. They will tell you. The key is to talk to the right people and ask the right questions. We will cover these topics in subsequent blogs.

2. Listen and Take Notes

It is one thing to ask questions. It is another thing to hear and absorb the answers. Most beginning analysts do not have great business sense. They may be well-trained on the latest machine learning algorithms, but they often fail to comprehend the most basic business concepts. Unfortunately, that business context is critical for success.

Sound familiar? If so, do not worry. Most new analysts start in the same place because Data Science programs rarely teach the business fundamentals. People learn on the job. Fortunately, you can do something simple to speed up the learning curve: take good notes. By good notes, I mean detailed summaries of the client’s responses that will allow you to really remember the conversation.

People often need to hear things a couple of times for the information to sink in. This is particularly true in new subject areas. What seemed obvious during a meeting may not be so obvious the next day. Crucial details or context may vanish. Good notes allow you to ‘repeat’ the conversation, recalling nuances that may otherwise evaporate.

3. Summarize and Validate

Summarizing and validating the conversation is the final best practice for ensuring you understand the business problem. This practice serves three functions.

a. It is an opportunity to restate the key points to further solidify your understanding.

b. This practice ensures you captured everything correctly, reducing the likelihood of basic misunderstandings that can derail your project down the line.

c. Finally, it provides your client with an opportunity to clarify and/or expand the conversation. It is difficult to cover everything in a single conversation. A follow-up email gives the client a chance to add information they may have overlooked.

Knowledge of algorithms is only one of many skills successful data scientists must master. In most real-world applications, this knowledge is arguably one of the least important factors in project success. Understanding the business problem and its operational context is generally far more important. After all, many algorithms can be used to solve most problems, but even the most sophisticated math will not save a solution based on a flawed business understanding.

Most data scientists start their careers without strong subject matter expertise in the business they support. Fortunately, they can overcome this limitation and accelerate the learning curve by acting like anthropologists. Interviewing clients, taking careful notes and publishing the results, are tried and true methods for ensuring every project gets off to the right start.

RELATED POSTS