When working with data, it’s easy to fall victim to analysis paralysis.
Rather than get down to the task at hand, we can get sidetracked by discussions, such as whether we should impute the mean or mode of a given dataset when handling missing data.
As data folks, we can analyze all the data we want, but if we have no way of translating the results into something meaningful, there’s no point in doing the work in the first place.
We all know that we work to answer questions or discover insights into a dataset, but how does one actually do that in the most effective way possible?
Context, Meet Relativity
Professor Dan Ariely is a world-renowned expert in behavioral science (who also happens to hold two PhDs). In his book, Predictably Irrational: The Hidden Forces That Shape Our Decisions, he notes that “humans rarely choose things in absolute terms.” Rather, he writes, “[humans] focus on the relative advance of one thing over another, and estimate value accordingly.”
In simple terms, this means that almost everything we do, we do in relation to something else.
That concept illustrates the importance of context, not only when it comes to decision-making but also for being an effective “searcher” of insight in a given dataset (otherwise known as the day-to-day task of a data scientist).
I work under the assumption that, without first having context about the data that you’re working with, no amount of academic degrees, computing power, or machine-learning algorithms can save you.
The Role of Analytics Translators
As a McKinsey group wrote in the Harvard Business Review, the search to fill roles in any data team primarily focuses on data scientists. But is that focus healthy?
The article makes the point that perhaps the most important person in any data team is the analytics translator:
And the role of those translators? To provide context. Analytics translators ensure that the efforts expended by the entire data team are guided by the core questions, use cases, and circumstances that surround the business. Building upon Ariely’s explanation, the translator helps us guide the efforts of our work relative to the larger picture.
Businesses are complex creatures. As data scientists, our role is to provide answers to ambiguous questions. We must simply do our best to present the best solutions — and the only way to do that is to have a full understanding of context.
In that respect, I’d say that context is more valuable than data. However, building context related to a dataset can be harder than it seems.
The Week’s Top Five:
The Best Data Scientists Get Out and Talk to People (Text)
As a data scientist, writing code and understanding the math behind a specific machine-learning model is simply not enough. “Great data scientists are deeply curious about the data and everything surrounding it.”
The #1 Way to Develop Context (Text)
TL;DR: “I mean, there is just no way to create the context without straight up DOING it.”
The 3 Kinds of Context: Machine Learning and the Art of the Frame (Text)
Jumping from one project to the next not only requires an ability to mentally shift gears in terms of tools and techniques, it requires something much harder: Your ability to shift context.
How to Improve Customer Experience Research by Understanding Your Users (Text)
A big part of building context is by watching what people actually do as opposed to what they say they do. This isn’t because they’re dishonest — it’s just a part of human nature.
With Big Data, Context Is a Big Issue (Text)
Data is practically useless unless it’s viewed within the context of a bigger picture. Arguably, context is how you move up the DIKW pyramid.
Like What You See?
If this issue of Missing Data has brought you any value, it would mean the world to me if you shared it with one friend using the buttons below. Thanks!
Grant