Data Analytics Will Never Be the Same

I've been working with spreadsheets since what feels like the beginning of digital time. My journey with spreadsheets began with VisiCalc around 1981. The interface was crude by today's standards, and no one had any idea what it would evolve into. But the ability to manipulate data—quickly adding up rows and columns of numbers back then—was incredible. I dabbled with Lotus 1-2-3 for a while in the mid-80s before finally settling into Excel around 1990. 

It's been quite the adventure, and I've developed a deep appreciation for Excel over the years. While I use Google Sheets for many tasks nowadays, Excel remains my go-to tool when starting a new analysis - it's wired into my brain.   Or at least it was.

On July 6th, 2023 OpenAI launched Code Interpreter to the world.  It wasn't the best name; shortly after that, it was rebranded as Advanced Data Analysis (ADA).  The moniker "ADA" probably honors Ada Lovelace, a trailblazing computer science luminary widely credited as the world's inaugural programmer.  The name, however, isn't essential – it's the capability that OpenAI unleashed on the world. 

ADA extended ChatGPT's capabilities by allowing real-time data manipulation and analysis, leveraging Python and other tools within ChatGPT. This dramatically enhanced ChatGPT's usefulness and opened a large number of use cases for analytic users. 

ADA allowed ChatGPT to become your personal data analyst—a "data scientist" in your pocket. I used this slide a lot when talking with clients over the past year about how GenAI tools could be used in their businesses. ChatGPT set the bar with ADA and has continued to evolve it. 

After spending a huge chunk of my career crunching numbers, I instantly saw the value in what ChatGPT could do, and I have been using it ever since. Let me give you three examples I've used in the last year to show you how you can use it today.

Example #1:  Data clean-up

If you've ever worked in a corporate environment or asked anyone for a dataset on something, the odds of it being perfectly formatted for easy analysis are essentially zero. I've worked with many organizations that use different CRM tools (Salesforce, Dynamics, Hubspot), and while you think people would enforce consistency on things like state names or phone number formats, it's rare that I get something I can use the first time.

Case in point – I had a client that gave me a data dump of over 5,000 phone numbers.  Ideally, for the work we were doing, we wanted the data in the format (XXX) XXX-XXXX.  The data looked something like this:

While I could probably have sorted the data, found a bunch of rules, and cleaned it up, it was much easier to prompt ChatGPT with the spreadsheet and a smart prompt and let it do the work for me. 

I used the prompt:

That's it – that's the whole prompt.  No giving it special rules or trying to identify every edge case – it knew what to do.  That single prompt saved me hours and hours of work.  What I got back looked something like this (see screenshot).  It formatted everything perfectly the first time through. 

After that I said "give me a link to download this data as an excel file" and I was off to the races.

Example #2:  Automated Trend Analysis

I can't even tell you how many times a senior executive sent me a spreadsheet with a note something like, "here's a total dump of the win-loss records in salesforce – see what trends you can identify and create a presentation that I can use tomorrow with the senior leadership team."   A note like that would stop all of the other work as a small team worked to decipher what was in the spreadsheet, what we could learn, figure out what mattered and then create some key talking points for the executive to use the next day.

With code interpreter – life got infinitely easier.  I literally used this prompt:

I wanted to see what code interpreter could learn on its own without me giving it too much direction and guidance.  

It gave me a quick summary of the data it found and then a plan for analysis:

I also got back this:

This is a great starting point for the type of analysis I want to do. I'm using a very simplified example here to show you how it works, but I've done this with spreadsheets with multiple thousands of rows of data and dozens to hundreds of columns.

It also created a nice number of starting charts for me to start analyzing the data as well:

Even if these aren't the exact starting points I would have chosen, they give me a great way to think about this, and the time savings are absolutely immense. 

Example #3:  Creating and manipulating charts from data

A picture is worth 1,000 words. I would argue that a well-created graph is worth 1,000 times that as well. ChatGPT's Code Interpreter is great at creating graphs and allowing you to manipulate them.  

Let's say you have a spreadsheet of credit card transaction data; you've uploaded it into ChatGPT and had it summarize the results – you might get something like this:

Let's say the data looks right – you can ask ChatGPT to "create a bar chart of this data," and you'll get something like this:

That's good—but not ideal—in terms of presenting the data to drive easy conclusions. A better way of presenting the data would be to say:

The result will look something like this:

These examples are just scratching the surface of what code interpreter can do. 

CONCLUDING THOUGHTS

The introduction of Advanced Data Analysis (ADA) in ChatGPT is changing the game for data analytics. 

These examples show that ADA makes it easy to clean up data, analyze trends, and create clear visualizations, all through simple prompts. It's bringing data science to everyone, giving users the tools of a skilled analyst without the complexity. The time-saving benefits and ability to generate quick, actionable insights are enormous for businesses and individuals. 

With this new era of data analytics on the horizon, the possibilities are endless. Whether you're a data pro or just starting, now's the time to dive into these powerful tools and see how they can transform your approach.