A Study by Volantsys Analytics & George Brown College, Part 2: Methodologies

Welcome back to our continuing series on Volantsys Analytics’s recent study on data science methodologies and improvements, in partnership with a team from George Brown College. In Part 1 of this series, we laid out the importance of data science and big data, introduced our research methodology, and examined the features of our survey respondents. In Part 2, we will dive deeper into our survey responses, focusing on the questions that talk about various data science approaches and implementation.

Data Science Methodology and Approaches

Survey respondent preferences on the organization and approach of data science and machine learning teams were fairly mixed. Participant preferences were almost equally distributed between decentralized teams that each serve an interest and centralized teams that span across multiple parts of the enterprise. The majority of respondents (65%) also favored a problem-solving approach that combined both top-down as well as bottom-up thinking, so no individual approach was a clear answer for general application. Based on these indeterminate responses, it seems that team structures and approaches depend heavily on project type, organization goals, and management styles as a whole.

Getting into the nature of the projects, respondents usually favored designating the task of scope definition to subject matter experts, product owners, and data scientists. At the same time, a diversified team consisting also of data engineers, data architects, and data and business analysts was deemed most efficient and contributory to success. Interestingly, project managers and product owners were seen as less necessary components of the team.

When trying to complete the project, Tableau was the most preferred data science analytical tool (65% of respondents), followed by AWS and AZURE cloud platforms (both 47%). On the other hand, the greatest challenge to data science project completion was poor data quality (76%) by far. Management issues, lack of leadership support, or culture issues were the main challenges to only 41% of respondents in comparison. This result is no surprise and reiterates the criticality of proper data management and governance.

 

That’s all for Part 2 of our series! Join us in Part 3 as we explore the questions that address the roles and responsibilities of data science teams.