JMP Pivot Chart: How to Create One Step by Step

To make a pivot chart in JMP, reshape and summarize your source data with Tables > Pivot, then open Graph > Graph Builder and assign the pivoted categories and values to the chart roles. For example, pivot monthly sales by product, then plot month on the X axis and each product’s sales as a series. Start with one observation per row and clear column headings; exact dialog options can vary slightly by JMP version.

What You’ll Need

A JMP data table with column headers and one record per row. Avoid merged cells, blank header rows, and subtotal rows.
A category column to group rows, such as Month or Region, and a category to spread across the pivot, such as Product.
A numeric measure to chart, such as Sales, Quantity, or Cost. Check that it is stored as numeric and has the appropriate continuous modeling type.
JMP with the Pivot and Graph Builder platforms. The menu path below is for JMP on desktop; names and dialog details may differ slightly between releases.

Step-by-Step

Check and prepare the source table. Open the data table and confirm that each column has a descriptive name and each row represents one record. For example, a sales table could have Month, Region, Product, and Sales. Make sure Month, Region, and Product are categorical columns, and Sales is numeric. In JMP, a column’s modeling type and its data type are different settings: a numeric month code may need a categorical modeling type, while a measurement such as Sales should generally be continuous. Use the column information or column properties to correct misclassified columns before pivoting.
Decide what each part of the pivot should represent. Choose the variable that should define the chart’s horizontal groups, the variable whose values should become separate pivot columns, and the numeric measure to summarize. For a monthly product chart, Month defines the rows, Product becomes the separate columns or series, and Sales is the measure. Write down this mapping before opening the dialog; it helps prevent swapping the two category roles.
Open the Pivot platform. With the source data table active, choose Tables > Pivot. In the Pivot dialog, select the source columns and assign them to the appropriate roles. In JMP releases that show role buttons such as Group, Label, and Data, use Group for the category that identifies output rows, Label for the category that creates the separate pivoted columns, and Data for the numeric measure. If your release presents the roles or controls differently, follow the dialog’s descriptions and look for the same three concepts: rows, column labels, and values.
Set how repeated records are summarized. If the source has several rows for the same Month and Product, JMP needs a summary statistic for their Sales values. Select a suitable statistic, such as Sum for total sales or Mean for average sales, when the dialog offers that option. Do not assume Sum is always right: use the statistic that matches what the chart should communicate. If each Month–Product combination is already unique, the result should not change the underlying values, but still check the output.
Create the pivoted table. Review the role assignments and any options for output names or missing values, then run the platform. JMP creates a reshaped table in which the grouping category appears in rows and the label category is represented by separate measure columns. In the example, the output may have a Month column plus columns for products such as Basic, Premium, and Deluxe. The exact column names depend on the source values and JMP’s naming rules.
Inspect the pivot result before charting. Confirm that the expected categories appear and that a few values match the source records or a manual total. Check whether missing combinations are blank or represented by a missing value. A blank may mean that no records existed for that combination; it does not necessarily mean zero sales. Decide whether that distinction matters for your chart. If a category is unexpectedly absent, check for trailing spaces, inconsistent spelling, or missing values in the source column.
Open Graph Builder. Make the pivoted table active and choose Graph > Graph Builder. Graph Builder opens a chart canvas with column roles around it. If the pivot output is not the active table, switch to it first so the chart uses the intended columns. You can also open Graph Builder and select the table from the available data-table controls, depending on your JMP release.
Assign the horizontal category and measure columns. Drag the grouping column, such as Month, to the X role. Then drag the pivoted numeric columns, such as Basic, Premium, and Deluxe, to the Y role. Depending on the graph type and JMP version, you can add multiple Y columns to compare the series in one graph. If the columns do not behave as numeric measures, inspect their data types and modeling types in the table; pivoted values should be numeric for a quantitative chart.
Choose a chart element that fits the question. Select a line element to show how values change over an ordered time axis, bars to compare amounts across categories, or points to show individual values. Graph Builder’s element controls are displayed on the platform. For monthly trends, a line chart is often a sensible starting point; for comparing totals by month, grouped bars may be clearer. Use the axis labels and legend to confirm that the plotted columns correspond to the products you intended.
Improve the chart and verify its meaning. Add a descriptive title, check that the X-axis order is chronological rather than alphabetical, and confirm the Y-axis scale and units. If the pivoted columns are plotted as separate series, make sure the legend identifies them clearly. Look for missing months, unexpectedly large totals, or a scale that hides smaller values. Compare at least one plotted point with the pivot table before sharing the chart.
Save or share the result. Save the JMP data table and the Graph Builder report using your normal JMP workflow. If another person needs to reproduce the analysis, retain the source table as well as the pivoted table and report, and give the columns meaningful names. A chart copied or exported without its underlying data may be harder to update when new records arrive.

Example: Suppose the raw table contains repeated transactions by Month and Product. Assign Month to Group, Product to Label, and Sales to Data, using Sum if the question is total monthly sales. The resulting table has one row per month and a separate sales column for each product. Plot Month on X and the product columns on Y. A line chart then compares the monthly sales patterns; a bar chart instead emphasizes differences in amounts at each month.

Common Mistakes

Reversing Group and Label. The Group role determines which values identify output rows; Label determines which values create separate columns. Reversing them produces a pivot in the wrong orientation. If the output has products down the rows when you expected months, change the role assignments and rerun it.
Using the wrong statistic. Summing a column that represents an average, rate, or already-aggregated value can give a misleading result. Select a statistic that matches the meaning of the measure, and check whether the raw rows are transactions or precomputed summaries.
Leaving numeric categories continuous. A numeric identifier such as 1, 2, and 3 may be treated as a measured quantity. Set category columns to an appropriate categorical modeling type so JMP does not treat the codes as a continuous scale.
Assuming blanks mean zero. A missing pivot cell can mean that no source record existed for the combination. Treating it as zero changes the interpretation. Check the source data and decide whether missing combinations should remain missing or be represented as zero for your specific analysis.
Charting the wrong table or wrong columns. After pivoting, it is easy to leave the original table active or drag a text label to Y. Confirm the active table, the plotted column names, and the values before interpreting the chart.
Ignoring category order. Text months can be sorted alphabetically, so April may appear before February. Use a date or suitable ordered category, or set the display order deliberately, then check the axis.

Tips

Keep a copy of the original data table. Pivoting creates a reshaped result, but the source is the best place to correct spelling, missing values, and data-type problems.
Use a simple test first: pivot a few rows or a small subset and compare the output with hand-checked records. This catches role and statistic errors before you build a polished report.
Choose the summary statistic based on the question. Sum answers “how much in total?” Mean answers “what was the average?” Count answers “how many records?” A count is not a substitute for a numeric measure such as revenue.
For time series, use a proper date column where possible rather than text month names. Dates make ordering and interval interpretation more reliable.
If the chart is crowded, plot fewer categories, use a meaningful filter, or separate series into panels. Adding every available category to one legend can make a correct chart difficult to read.
When you need the categories in a legend or color role rather than as multiple Y columns, consider reshaping the pivot output back to a long format with Tables > Stack. Stack the measure columns into a value column and retain the source column names as a label column; then put the category on Color in Graph Builder. Check the Stack dialog’s roles in your version before running it.
Save the Graph Builder script or report with the table when you expect to revisit the analysis. Keeping the chart linked to a clearly named data table makes it easier to reproduce and update.

Troubleshooting

Tables > Pivot is unavailable: Confirm that a JMP data table is open and active, and check that you are using a JMP desktop release that includes the Pivot platform. Menu availability can depend on the product and version. If you cannot use Pivot, Tables > Summary can summarize data by groups, though its output is not necessarily the same wide layout.
The output has more rows or columns than expected: Inspect the source category values for spelling differences, extra spaces, inconsistent capitalization, and missing entries. Each distinct category can produce a separate group or pivot column. Correct or standardize those values in the source, then rerun Pivot.
The totals seem too large: Check whether repeated records are being summed intentionally, whether the measure was already aggregated, and whether duplicated source rows exist. Compare the pivot result for one category with a filtered view of the source table.
Graph Builder shows no useful chart: Verify that at least one numeric measure is assigned to Y and a meaningful category is assigned to X. If the pivoted measure columns are character data, correct the source column’s data type and recreate the pivot. If there is no variation in the plotted values, try a bar or point element and inspect the table directly.
Only one series appears: Add the other pivoted numeric columns to Y, or use Tables > Stack to create a long table with a value column and a category-label column. In the long layout, place the value column on Y and the label column in Graph Builder’s Color role.
Months or categories appear in the wrong order: Check whether the source values are text. Use an actual date or set a deliberate value order for categorical data, then rebuild or refresh the graph. Do not rely on alphabetical order to represent time.
Some combinations are missing: Check whether source records exist for those category pairs. If they do not, the pivot may correctly leave those cells missing. Decide how the absence should be displayed before interpreting the chart as zero.

Frequently Asked Questions

Is a JMP pivot chart a separate chart type?
Not usually. Pivot describes how the data is rearranged into a summary layout; Graph Builder is where you choose the chart element, such as a line, bar, or point chart.

Can I make a chart without creating a pivot table first?
Yes. If your data is already in a suitable layout, open Graph > Graph Builder and assign the category and measure columns directly. Pivoting is useful when you need to summarize records or compare categories that should become separate series.

Should I use Sum or Mean in the pivot?
Use the statistic that answers your question and fits the meaning of the source data. Sum is appropriate for totals from transaction-level records; Mean is useful for averages. Check how the source measure was created before choosing.

Can I include more than one category in Group?
Yes, when you need output rows defined by a combination, such as Month and Region. The result will have a row for each distinct combination. More grouping columns can create a larger table, so include only the dimensions needed for the chart.

How do I show pivoted categories as chart colors?
Use the long-data layout: stack the separate measure columns into one value column while retaining their former column names as labels. In Graph Builder, assign the value to Y and the label to Color. This is often easier to manage when the number of categories changes.

Will the chart update automatically when I add new source rows?
Do not assume that adding rows to the original table will automatically rebuild a separate pivot result. Rerun or update the pivot and check the Graph Builder report after changes; the exact behavior depends on how the tables and scripts were created in your JMP version.

Try it: the Percentage Formula Generator: percent of, percent change, increase and decrease, with the formula.

More in Charts & pivot tables

Sources: Function behaviour and menu paths: Microsoft Support documentation for Excel. Method & sources

Spotted an error in this guide? Let us know.