Saturday, February 16, 2013

Know Your Analytics Professionals: Tool Users vs. Coders

Data work falls into a number of different subcategories. One of these subcategories is analytics, also known as business intelligence, data mining, or machine learning. While these terms can be distinguished from one another based on nuance, they all have in common the use of computer programs to find patterns in data that can be generalized for reporting, optimization or prediction.

Analytics professionals fall into two categories, tool users and coders.

Tool users generally live in Stata, SPSS, SAS or sometimes Excel. They can have a reputation in industry for being on average less statistically literate than coders, but can also be more productive by virtue of their streamlined analytical methods that have been honed and optimized over the years by tool manufacturers. While they can be efficient, they tend to be disconnected from overall business processes. These analysts typically need to have data extracted for them from a central system. They do their analysis offline and usually work on aged data.

Coders use R, Python or sometimes SQL. They tend to be less concerned with standard approaches or productivity, and more concerned with flexibility and data integration. While they can be less efficient than tool users when working with extracted, disconnected data, coders will generally have a better sense of how a given process fits into the big picture, and they are more likely than tool users to be able to bolt their work to a larger system and access fresher, sometimes real-time data in an automated way. When this happens they can eclipse the efficiency of packaged tools and provide extraordinary additional value to an analytics effort.

These are of course generalizations. Nonetheless, a data or an analytics manager should be mindful of these distinctions when adding someone to a team. Balance is ideal. It may seem that in a shop where SAS is the standard, you might not want a Python developer lurching around. However, the ability to take proven processes into a live production environment and turn last-quarter’s reports into yesterday’s dashboard might be just the thing to take your team to the next level.

No comments:

Post a Comment