Data Dunkers - LLM Data Analysis

LLM Data Analysis

Large Language Models (LLMs) like ChatGPT are powerful AI tools that can also read, describe, and analyze data. In this activity, you will download a set of measurements from the Your Data - Measurements activity and upload it to an LLM for exploration.

Step 1: Download the Data

Click this link to download the basketball measurements as a CSV file.

Step 2: Choose an LLM

Here are some AI tools you may want to use. No login is required, but you may need to check with a teacher or guardian before using them. Avoid sharing any personal information.

Google Gemini

Google's AI assistant. Supports file uploads directly in the chat.

Perplexity AI

Great for data summaries and questions.

Microsoft Copilot

Microsoft's AI assistant. Supports file uploads and can summarize and analyze CSV data.

Step 3: Upload the CSV and Start Asking Questions

Once you've opened one of the tools above, look for a paperclip icon, an upload button, or a "+" button to upload the CSV file you downloaded.

Then try some of the example prompts below.

Example Prompts to Try

Understanding the Data

Summarize this dataset. What columns are included and what do they represent?

How many rows are in this dataset? Are there any missing values?

Generating Visualizations

Create a scatter plot of height vs. wingspan. What trend do you notice?

Make a bar chart showing the average value for each numeric column.

Plot a histogram of the height column. Describe the distribution.

Finding Correlations

Which two columns are most strongly correlated? How do you know?

Is there a correlation between height and reaction time? Show a chart and explain your reasoning.

Generate a correlation matrix for all numeric columns and highlight the strongest relationships.

Deeper Analysis

Are there any outliers in the data? Which rows are they and how do they compare to the rest?

What is the average, median, and standard deviation for each numeric column?

If you were a coach, what insights from this data would you use to build a better team?

Tips for Getting Better Results

Be specific: Instead of "analyze the data," ask "compare the average height between groups."
Ask follow-up questions: If the first answer isn't clear, ask the LLM to explain or try a different approach.
Challenge the AI: Ask it to explain why two variables might be correlated, not just whether they are.
Check its work: AI can make mistakes. Compare its answers to what you already know about the data.