It’s called science for a reason. Data science requires questions, hypotheses, discovery, analysis, and recalibration.  While the words I’m using to describe a data science are different than software development, the process and mindset are very applicable to agile methodologies.

And just like software development, data science requires a multi-disciplinary team. Data science is not just about the machine learning models and data visualizations, it must also include all the dataops and proactive data governance to cement an end to end journey.

And just like there’s technical debt, there’s data debt, including the need to iteratively improve data quality, update data catalogs, and move downstream analytics back into centralized databases, data warehouses, and data lakes?

Data Science Requires Balanced Priorities

I have several articles on applying agile in data science. My first one was on using agile to find value in dark data, and my most recent one covered three ways to apply agile in data science and dataops.

Successful data science and becoming a data-driven organization requires a multi-disciplinary team to balance priorities. If you overly invest in data science, you’re likely underinvesting in dataops and data governance and probably r.eating a mound of data debt. Similarly, pushing too hard on dataops and data governance, and you’ll lose stakeholder interest if they don’t see the value.

Here’s more in Episode 14 of 5 Minutes with @NYIke on Agile Data Science.