GPT-4 Excels At Data Analysis And Visualizations Through Its Code Interpreter, Users Find

The list of professions that ChatGPT could disrupt only grows longer.

After writers, coders, consultants, artists and doctors, GPT-4 seems set to take over much of the work performed by data analysts. The functionality was recently revealed by OpenAI President Greg Brockman in a TED talk, where GPT-4 was able to understand a csv file, figure out what the columns meant, suggest what sort of data analysis to do, and even create graphs and charts.

ChatGPT’s data interpreter plugin allows the LLM to run code, just as a data scientist would. As users have had a chance to play with the code interpreter, they’re managing to perform some interesting data analysis of their own. A user merely uploaded the crime data in San Francisco, and GPT-4 was able to come up with data analysis ideas, plot graphs, and even visualise trends.

The code interpreter feature on ChatGPT is the most mind blowing thing I've seen yet.

All I did was upload a CSV of SF crime data and ask it to visualize trends(!!) pic.twitter.com/pkFdPqgAzb

— John Backus (@backus) April 29, 2023

Another academic was able to similarly get ChatGPT “Fun case of GPT Code Interpreter coming up with neat solutions: I gave it a list of lighthouses in the US by latitude & longitude, and one of Frank Loyd Wright houses by street address I asked it to identify places where both structures were less than 10 miles apart. Did it well,” professor Ethan Mollick tweeted.

Fun case of GPT Code Interpreter coming up with neat solutions: I gave it a list of lighthouses in the US by latitude & longitude, and one of Frank Loyd Wright houses by street address

I asked it to identify places where both structures were less than 10 miles apart. Did it well pic.twitter.com/0aZvM7sLHa

— Ethan Mollick (@emollick) May 1, 2023

Data analysis represents another jump in ability for GPT-4. All GPT-4 needs is some labeled data, and is able to immediately figure out the context, perform exploratory data analysis, and even create graphs and charts. All this is something a trained data analyst would’ve taken several hours, if not days to complete, but ChatGPT seems to does it effortless in seconds. It’s not as though this capabilities will make data analysts obsolete — there still might be people required to udnerstand and make sense of the data — but GPT-4 will, like with other professions, require data analysts to add value above and beyond what GPT-4 can do instantly and virtually for free.