By: Bill Baez, PhD, Vice President-Strategy, Ascend Innovations

Data scientists begin analysis by working to understand the people and data involved before turning any numbers into actionable, data-driven insights. Data science frameworks often begin with an initial step of “Business Understanding”, as seen in the Cross-Industry Standard Process for Data Mining and Microsoft’s Team Data Science Process. However, like many analytical fields, data science is often defined using a cold, precise tone, like this one from a recent CIO article:

“The goal of data science is to construct the means for extracting business-focused insights from data. This requires an understanding of how value and information flows in a business, and the ability to use that understanding to identify business opportunities.”^[1]

And while true, definitions like these often gloss over a fundamental aspect of these data-driven insights – the people using them.

While data science frameworks often focus on the need to understand the business, they lack an emphasis on the individuals faced with problems that require the data scientist’s expertise to solve. To provide great and effective solutions, data scientists must be willing to understand what their users feel about their problems. Data scientists need to first empathize with their clients.

A growing trend within data science is incorporating elements of design thinking into data science frameworks. Design thinking has long been used in product development and considers empathy to be the first step where researchers can get a better understanding of the problem users are trying to solve.

Data science is fundamentally a collaborative effort between you and those using your solution. Data scientists that foster a deep interest in understanding the people for whom their products are built, create more effective data-driven products and services than those that deliver a sound technical package. Data scientists must place the same weight on understanding their user’s needs as they do in feature engineering or picking the right machine learning model. Not seeing the problem through the user’s eyes can lead to weeks, if not months, of wasted effort creating a model or dashboard that is ultimately not used.

Historically, data scientists have focused on the technical aspects of a project to improve performance. Improved accuracy is only part of the equation when examining a product’s effectiveness. Increasing a model’s accuracy from 80% to 83% isn’t always the right metric to measure its impact on the problem you set out to solve. You want to find out how often that model or dashboard is being used and in what context. You also want to understand how much the decisions made by the models are acceptable to users, how to build trust in the results, and how your users identify value from your product. The answers to these questions will help data scientists develop solutions that are not only technically right but also effectively right for the people using them.  

Bill Baez, PhD, Vice President-Strategy, Ascend Innovations

Bill is currently the VP of Strategy at Ascend Innovations in Dayton, OH. In his role, he works closely with multiple departments to provide socially impactful, data-driven products and services to organizations trying to solve complex community problems.

^[1] https://www.cio.com/article/221871/what-is-data-science-a-method-for-turning-data-into-value.html