Exploratory vs. Explanatory Analysis
When I was working with a company last week I found myself talking about the difference between Exploratory vs. Explanatory Data Analysis and thought it would be a good topic for a blog post.
Exploratory Data Analysis
Exploratory Data Analysis, is when you’re looking at a fresh batch of data and trying to:
Check it to see if the data is correct, complete and consistent.
Figure out what useful (actionable) insights you can get out of it.
Because it’s your first draft - you don’t want to create something you love. Hold off on the fancy formatting here.
For data professionals Exploratory Data Analysis is a really big deal, sometimes it’s even shortened to it’s three letter acronym EDA. “I’m gonna go run some EDA on that big data set you sent over”.
Non Data Professionals sometimes skip this step to their detriment!
This can go poorly two ways :
(Better) Your great insights get mixed in with a lot of unclear data making it hard for your key points hard to find and interpret.
(Worse) All the data you’re working with is wrong and you don’t know it.
The best way to do Exploratory Data Analysis is in the tool you’re most comfortable in, in the most disposable way possible. You are checking to see what works and what you want to discard. I sometimes call this the “red crayon draft” because I’d never turn in anything I wrote in red crayon.
Explanatory/Persuasive Data Analysis
You’re ready for this step AFTER you’ve done your EDA.
At a minimum, this should be in a new window, tab, slide deck or even piece of paper.
Data is a form of communication.
You know what you want to communicate and you’ve spent time and energy tailoring it to your audience and venue.
By clearly distinguishing between Exploratory Data Analysis and Explanatory Data Analysis, you enhance the quality of your data communication and your audience receives a well-structured, impactful narrative that inspires action.
Try it, or reach out to see how I can help you and your team learn to do this effectively.
My Cat: Is Fantastic
Meet Inertia:
She’s cute and fuzzy but also a source of professional inspiration. She’ll sometimes sit on me till I get something hard done 😻.
Data Therapist
Working with data is an emotional journey.
Often a manager says “Go get me some data” or “Let’s make data driven decisions”, as if employees are going to go to Dataland to pick up some insights and return with a polished slide deck before 5pm.
Before people can begin working with data there’s a lot of emotional baggage that they have to deal with. I hear things like:
“I’m not good at data.”
“I was never good at math.”
“I was an English/French/Theater/Other Liberal Arts major in college.”
These are 100% real comments I’ve gotten from talented professionals - experts in their chosen field.
Before I can tackle any data problem - I need to tackle a self image problem. Someone warning me that they “Majored in English” is them admitting that despite impressive professional credentials, data still makes them nervous.
While I work with adults, these negative math/data perceptions are formed EARLY, by 2nd or 3rd grade and, not surprisingly (😢), disproportionally affect women and girls.
I can help overcome some of this by being friendly, using approachable language, animated presentations, lots of praise. But, I have found, the thing that works best by far is the “just one thing” approach.
You don’t have to be a data wizard, a math genius or a computer science major. I promise I’m not teaching anyone multivariate calculus. I’m not going to ask you to submit your answers in Python or Java Script. Right now we’re going to push one button, learn one concept go one step at a time.
I usually can’t shift someone’s whole self image that is bound up with their Musical Theatre degree but we can make incremental progress every day and bask in the glow of accomplishment that progress imparts. 😁