Exploring with AI
Last updated
Last updated
Large language models (LLMs) like ChatGPT offer several different features to support exploratory (and explanatory) data visualization.
First, LLMs can analyze visualizations, providing a powerful way to bridge the gap between raw data and actionable insights. LLMs can assist you in interpreting visualizations, explaining key trends, finding patterns, and identifying anomalies. For instance, ChatGPT can identify the overall direction of a time series chart, highlight outliers, or explain relationships between variables shown in a scatter plot. These features are particularly helpful as you get started in data analysis or if you want a quick summary of the visualization. By describing visual patterns in plain language, LLMs enhance accessibility and make data more understandable for diverse audiences.
Second, LLMs can suggest new visualizations. For example, ChatGPT can suggest appropriate chart types based on the data and analysis goals, such as recommending a scatter plot for examining relationships or a bar chart for comparing categories. As you learn more about data visualization, LLMs can act as a resource and a check.
Third, ChatGPT can also assist with coding, providing example scripts in visualization libraries like Plotly, Matplotlib, or Seaborn to generate specific charts in Python. Relatedly, LLMs can suggest advanced visualization techniques such as using subplots, animations, or annotations to enhance interactivity and storytelling. This capability makes tools like ChatGPT valuable not only for analyzing visualizations but also for empowering users to craft compelling and effective visual narratives.
Go to ChatGPT (chatgpt.com). If you prefer to use another LLM, that is fine as long as it has multi-modal capabilities. Multi-modal means that the LLM can process images as well as text. We will prompt ChatGPT with both text and the created image to analyze the visualization.
Here is what ChatGPT said about the above visualization using the following prompt:
The key takeaways from this visualization of Boeing's stock price for 2024 are:
Overall Trend: The stock price generally declined throughout most of 2024, starting above $240 and dropping below $150 at its lowest point.
Early Decline: There was a sharp drop in the stock price early in the year, indicating a significant event or market reaction during that period.
Mid-Year Stability: The price shows some level of stabilization or smaller fluctuations between May and July 2024, hovering around the $180 range.
Late-Year Recovery: Towards the end of the year, the stock price rebounded, rising from below $150 to approximately $180.
Volatility: The line chart shows considerable volatility throughout the year, with frequent ups and downs, indicating an uncertain market environment for Boeing.
This chart is useful for identifying general trends and periods of volatility but does not provide context for specific events affecting the stock price.
Initial Reflection: What are the ways to improve the analysis from ChatGPT?
Initial Reflection: If you are unhappy with the results, you can try a slightly different prompt such as What are the important patterns in this chart? You should try 2-3 different prompts and compare the results until you feel satisifed.
