GPT-4’s Prompting Effectiveness for Python Dashboards: Comparing Dash, Panel & Streamlit

Author(s): John Loewen, PhD

Originally published on Towards AI.

Prompting GPT-4 for multi-visual interactive dashboard creation

Top highlight

Dall-E 2 image: impressionist painting of 3 computer screens showing data visuals.

As a comp sci professor, over the past 6 months, I have heavily integrated GPT-4 into my data visual creation workflow.

I have found recently that GPT-4 has improved in leaps and bounds in its ability to create code for Plotly dash multi-visual dashboards.

This has piqued my curiosity on whether GPT-4 has also upped its game to provide seamless dashboard creation for other Python dashboard libraries.

Does it have the chops to handle Panel and Streamlit as well?

Let’s find out!

For this exercise, I will be using the downloaded file (saved as “happiness_years02.csv”), which includes data on global happiness metrics by country over several years (2015–2022).

It can be downloaded HERE.

To start with, we want to load the dataset with pandas, focusing on each country and the happiness score by year (2015–2022). To crystalize the view for GPT-4, we can ask GPT-4 to look at the dataset and tell us what it sees.

We can click the attachment icon in the GPT-4 main window and upload our dataset, then prompt GPT-4 for analysis:

This way, we can clarify any anomalies that may exist between GPT-4’s interpretation of the data vs. our interpretation.

GPT-4 Response:

Yep, this… Read the full blog for free on Medium.

Join thousands of data leaders on the AI newsletter. Join over 80,000 subscribers and keep up to date with the latest developments in AI. From research to projects and ideas. If you are building an AI startup, an AI-related product, or a service, we invite you to consider becoming a sponsor.

Published via Towards AI