Prepare data with Minitab AI
Supported data prep requests
- Clean or transform data using supported cleanup steps.
- Join two data sources into one dataset.
- Union up to four data sources.
- Transpose a selected dataset.
After Minitab AI updates the pipeline, you can review, modify, or undo the changes.
- Combine and Reshape Data
- Sort and Filter Data
- Change Column Properties
- Transform Column Values
- Remove Data
Access Minitab AI
. Minitab AI is not available with empty datasets or when the data pipeline has warnings or errors.
To stop generating steps, select 
.
Important
AI technology may make mistakes. It is your responsibility to ensure the output is accurate, appropriate, and meets your organization's standards and requirements. For more information, see the Minitab Trust Center.
Undo/Redo AI steps
You can edit or delete AI-generated steps. If Minitab AI produces multiple steps from the same prompt, you can undo or redo those steps together. Undo and redo apply only to the steps shown in the pane.
Tips and requirements
- The dataset must be open and must contain data
- If any prior preparation steps make the dataset invalid, then the AI is not available.
- Be as clear as possible in your prompts
- Prompts in English give the best results.
If the AI rate limit has been reached, you must wait a few moments before trying again.
- Use a series of smaller prompts instead of one complex prompt
- Minitab AI generates a maximum of 30 steps with one prompt. If a single prompt
generates more than 30 data prep steps, an error is issued and no steps
are generated.
A series of simple prompts can help improve efficiency and accuracy of the generated steps.
- Requests to Minitab AI are additive
- When Minitab AI prepares your data, it adds a new step or transformation
to the pipeline based on the current data. Existing pipeline steps
remain unchanged. To insert steps in a particular order, make sure the
step is selected in the Steps pane.
Note
The AI-generated steps can return errors in the Steps pane if they conflict with previous or subsequent steps.
- Provide feedback on the generated steps
- Give a thumbs up or thumbs down to provide anonymous feedback.
- Several data cleaning steps require that the data are in equal length columns
- When you have unequal columns and try to use the Sort, Filter, or Merge operations, the columns are populated with missing values to make column lengths equal.
Examples of AI prompts
| Goal | Example prompt | Result |
|---|---|---|
| Replace values | for interest in a hearing app, use only yes no or maybe in that column | Adds a step that standardizes the values. |
| Filter rows | keep only rows with maybe for interest in hearing app column | Keeps only the rows with Maybe in the Hearing App column. |
| Extract text values from a column | split the desired app features | Splits the original column into multiple new columns, using the comma as the delimiter. |
| Join two data sources into one dataset | join the customer data with the order data by customer ID | Adds a Join transformation to the pipeline. |
| Combine data sources | combine the January, February, and March sales data | Adds a Union transformation to the pipeline and matches the columns. |
| Transpose data | transpose this dataset | Adds a Transpose transformation to the pipeline. |
