Advanced Computing in the Age of AI | Thursday, March 28, 2024

Data Scientists: Avoid These Top 4 Pitfalls 

When working with business executives, the biggest pitfalls in data science boil down to errors in communication. For true success, data scientists must be bilingual to effectively accomplish their work. This means being equally versed in technological jargon as well as being able to communicate with business stakeholders. When we look at where data science fails, it’s usually in communications.

Data science is a relatively new field, first coined in 2001 by William Cleveland to “enlarge the major areas of technical work of the field of statistics.”

Today, the demand for data scientists is astronomical, because company data can offer tremendous value in edging out competitors—if properly utilized. Combined with someone who knows how to take that information and formulate a strategic business plan, you have a winning hand.

However, high demand for experienced talent in the field has stretched the candidate pool thin. Companies know hiring someone who has been around the block a few times will allow them to quickly see results.

These experts are well-versed technologically and understand business strategy, but this takes an immense amount of time in the field to acquire. There are four main reasons data scientists fail. Avoiding these pitfalls will ensure success in their roles and bring value to companies seeking to leverage data.

Not Using a Common Language

Data scientists must be bilingual in technology solutions and business strategy. Data scientists must have the technical acumen to discuss complex ideas with their teams, but also must know how to speak to a company stakeholder who focuses on business issues and may not have a technical background.

Using a common language between data scientists and executives will help create meaningful and informative discussions. Using terms like autoregression, backpropagation, kurtosis, mahout, tokenization, etc., will likely fall on deaf ears, but explaining the concepts as they relate to the broader business with anecdotal evidence will provide understanding to those looking at the larger picture.

Any misunderstanding in the terminology is a good sign there will be misalignment so there needs to be a core vocabulary utilized by all parties.  Business vocabulary is best and should be something in which all data scientists are trained. This ensures all parties understand how data science helps the company, and how a strong, strategic plan can be created without the risk of misunderstanding.

Insufficient Domain Expertise

Easily the worst pitfall on this list, insufficient domain expertise negates the value a data scientist can bring to their teams. Machine learning algorithms can only be properly utilized if those in charge of the information have a deep knowledge of the industry and the intricacies of the data points being used.

Domain expertise is a skill is not easily acquired and requires experience studying up on industry trends.

Kalev Leetaru, a contributor with Forbes, explains this pitfall very well:

“[A lack of domain expertise] creates a dangerous situation where data scientists are often unfamiliar with the nuances of the data they are working with or the assumptions of the domain they are working in and produce analyses that inadvertently lead their organizations astray.”

A team with a solid number of domain experts will bring significantly more value to the company and garner more actionable insights that be activated rapidly in specific markets.

Difficulty Understanding Business Goals

A company with a digital footprint accumulates mass amounts of information and it’s the role of data scientist to synthesize and strategize the best way to use that data to meet business goals. Difficulty in understanding the business goals, though, will produce a very poor strategic plan.

A data scientist must be able to take unstructured data, clean it up, and find actionable insights inside that information. If the data scientist doesn’t have a firm grasp on the business goals, there will be no actionable insights gleaned from the information and time spent analyzing and formulating the strategic plan will leave the larger business stagnant.

Circumventing this pitfall is not unattainable. Taking a business class, speaking with business leaders, and reading about your industry can provide a better understanding of the company and how to best use the data to achieve meaningful corporate goals.

Lack of Curiosity

Great data scientists are often drawn to the field because they are curious by nature. Curiosity is one of the most critical soft skills someone in this industry can have to be successful. Data science is not a career for someone seeking a comfortable 9-5 job crunching numbers all day. Most of the time is spent talking to people and asking questions to have a firm understanding of how their research will business success.

Curious people are more likely to want to educate themselves on a topic. Curiosity is a soft skill, yes, but it can be cultivated. Everyone is born curious. It is part of human nature. Setting aside time to learn more about a subject increases the chance for inspiration to strike. Inspiration is the root of curiosity.

Successful data scientists seek challenges and thrive in solving mysteries. Those people with technical and creative sides, will be the ones most sought after.

Conclusion

Addressing these top four potential pitfalls will also help inexperienced data scientists or those just entering the field bring value to their roles.  The industry is relatively new and the field frequently undergoes transformation and change. Keeping skills up-to-date is a sure way to advance your craft, career and opportunities.

Experience is not the only way to keep from falling into a career pit, as there is only so much that can be learned on the job. All data scientists can become stagnant, so it is important to foster that curiosity and continue to learn.

About the Author

Luming Wang is the Chief Technology Officer of AI and machine learning vendor, ElectrifAI, where he works to ensure that the company’s underlying technology makes machine learning practical and relevant for enterprise customers. Before joining ElectrifAI in February 2020, Wang served as the head of data for Millennium, a global alternative investment firm, as the head of deep learning at Uber, and in roles at Microsoft, Amazon, Cognos and CA Technologies. He is a graduate of the University of Science and Technology of China.

 

 

EnterpriseAI