ChatGPT - Data Analysis

ChatGPT can load a data file, run the analysis for you, and return results, charts, and cleaned files right in the chat. You describe what you want in plain English; you don't need to write code. This article explains how to use data analysis in University of Maine System (UMS) ChatGPT Edu.

Data analysis is part of a regular chat. On Instant it's free. Medium, High, and Extra High use 10 credits per message. See ChatGPT - Plans & Usage Limits.

How data analysis works

Behind the scenes, ChatGPT writes and runs Python code in a secure, sandboxed environment. When you upload a file:

1. ChatGPT examines your data. It reads the structure, including column names, data types, and the shape of the dataset. It may show an interactive table you can scroll through.
2. You ask a question. Something like "What is the average GPA by major?" is enough to begin.
3. ChatGPT writes and runs the code, then returns the results, tables, or charts.
4. You refine. Each follow-up builds on the last. Adjust filters, ask for a different chart, dig deeper, or export the results.

To see the code ChatGPT used, select View Analysis when it appears in a response. Use it to check the logic, learn the approach, or reuse the code elsewhere.

Get started

1. Open a new chat, or a chat in a Project.
2. Select the attachment icon, or drag and drop a data file into the chat.
3. Wait for ChatGPT to load the data.
4. Ask a question about the data in plain English.
5. Review the results and keep going with follow-up questions.

Supported file types

• CSV (.csv): the most reliable format for tabular data
• Excel (.xlsx, .xls)
• TSV (.tsv)
• JSON (.json) and XML (.xml)
• SQLite databases (.sqlite, .db)
• Parquet (.parquet)
• Text files (.txt) with structured data
• PDF: ChatGPT can pull tables out of PDFs, but results depend on how the PDF was made

For best results, use CSV or Excel files with clean headers and consistent formatting. Files can be up to 512 MB, but spreadsheets and CSVs are practically limited to about 50 MB, depending on row size. You can also pick files from connected apps that are approved in the UMS ChatGPT Edu workspace.

What you can do

Explore and summarize

Ask ChatGPT to describe a file. It can count rows and columns, list column names and types, find missing values, and give summary statistics.

Example prompts:

• "Describe this dataset. What columns are there, and what do they contain?"
• "How many rows have missing values? Which columns are affected?"
• "Give me summary statistics for all numeric columns."

Clean and transform

Remove duplicates, fill or drop missing values, rename columns, convert data types, filter rows, and restructure data.

Example prompts:

• "Remove any rows where the email column is blank."
• "Convert the date column from text to a proper date format."
• "Split the full_name column into first_name and last_name."
• "Pivot this data so each department is a column and each month is a row."

Calculate and aggregate

From simple sums and averages to grouped totals and cross-tabulations.

Example prompts:

• "What is the total revenue by region?"
• "Calculate the average response time grouped by ticket priority."
• "Show me the month-over-month percentage change in enrollment."

Run statistical analysis

Descriptive statistics, correlations, regressions, hypothesis tests, and related methods.

Example prompts:

• "Is there a correlation between study hours and GPA in this dataset?"
• "Run a linear regression with GPA as the dependent variable and study_hours and attendance_rate as independent variables."
• "Perform a t-test comparing average scores between Group A and Group B."

Create charts

Let ChatGPT pick a chart type, or ask for a specific one.

Interactive charts let you hover over data points and zoom. Ask ChatGPT to change colors, labels, or layout, and download charts as images for presentations or reports.

Example prompts:

• "Create a bar chart showing total headcount by department."
• "Show me a line chart of monthly ticket volume over the last two years."
• "Make a scatter plot of study_hours versus GPA, color-coded by major."
• "Generate a heat map of correlations between all numeric columns."

Work with tables

In the interactive table, you can scroll, select rows or columns, and ask about what you selected. You can also ask ChatGPT to create new tables, add calculated columns, filter or sort records, or restructure data. Tables download as CSV files.

Merge datasets

Upload more than one file and ChatGPT can join them on a shared identifier, such as a student ID, and answer questions across both.

Example prompts:

• "Merge these two files on student_id and show me students who appear in both."
• "Join the sales data with the product catalog on product_code and calculate revenue by category."

Ideas by role

• Researchers: Explore survey or experimental data with descriptive statistics, correlation matrices, and regressions before moving to specialized tools for final analysis.
• Staff and administrators: Analyze enrollment trends, budget data, Service Desk ticket volumes, HR metrics, or event attendance, and build charts for reports.
• Faculty: Look for patterns in grade distributions or assignment scores across sections, and create visuals for program reviews or accreditation reports.
• Students: Work through statistics assignments and class projects. Ask ChatGPT to explain each step to learn the method, not just the answer.

Tips for good results

Start with clean data. Clear column headers, consistent date formats, and no extra rows help ChatGPT read the file correctly.

Be specific. "Analyze this data" usually returns general summary statistics. A specific request gets a more useful result:

Show me the top 10 departments by average salary, sorted from highest to lowest, as a horizontal bar chart.

Stay in the same chat. Start broad, then refine filters, calculations, or charts with follow-ups.

Check important results. Use View Analysis to inspect the code. For decisions with real consequences, confirm key findings another way.

Use a Project for recurring analyses. Add instructions with your preferred format, metrics, and terms. See ChatGPT - Projects.

Need a finished report or spreadsheet? Faculty and staff can use Work for longer tasks that end in a finished file.

Save your work. Download charts, tables, and cleaned files before you delete the chat.

Limitations

• No live data connections unless an approved connected app is available in the UMS workspace. Export the data to a file and upload it instead.
• The environment is sandboxed. The code can't reach the internet or outside resources.
• Large files can hit limits. Filter to the records you need, or split the data into smaller files.
• Results need checking. ChatGPT can miscalculate, misread a column name, or pick the wrong statistical method.
• Charts aren't dashboards. They live in the chat and don't refresh on their own.

Know your data before you upload it. Follow UMS data governance rules. Don't upload sensitive or restricted data unless UMS policy explicitly allows it.

Questions about data analysis?

Start a chat Call 1-800-696-4357

Prefer to stop by in person? Check Service Desk Hours and Availability for current hours, phone, email, chat, and walk-up locations.

Environment

• University of Maine System (UMS) ChatGPT Edu workspace
• Data analysis
• Eligible UMS students, faculty, and staff
• Web browser at chatgpt.com, plus the ChatGPT desktop and mobile apps