Configuring two metrics causes SQL statements to appear in the echart legend configuration, while configuring one metric works normally.

[MidnightOakMapleNote]

Hello everyone, does anyone know why this situation occurs?

During chart configuration, if only one metric is configured, the configuration works normally. For example, if only "COUNT_DISTINCT(classification_name)" is configured, the legend will display each classification name from the table, which is [classification_name1, classification_name2.....] at this time.

However, if two metrics (COUNT_DISTINCT(classification_name) and COUNT_DISTINCT(custom_type)) are configured, as shown in the screenshot, the legend configuration of the entire echart will become ['COUNT_DISTINCT(custom_type) classification_name1', 'COUNT_DISTINCT(custom_type) classification_name2'], etc. Could this be a bug?

The expected result is that when two metrics are configured, the legend should be [custom_type1, classification_name2, custom_type2, classification_name2], without the SQL expressions included.

Thank you all for your help. I would be very grateful for your answers.

[dosubot[bot]]

This happens because Superset generates legend labels for echart visualizations by combining the metric display name (which defaults to the SQL expression or aggregate function if no custom label is set) with the groupby dimension value. When you use only one metric, the legend shows just the groupby value. With multiple metrics, the legend label includes both the metric and the dimension, so if your metric doesn't have a user-friendly label, you'll see the raw SQL expression in the legend.

This is expected behavior, not a bug. To get cleaner legend labels, set a custom "label" or "verbose name" for each metric in your chart configuration. Superset will use these for legend labels instead of the SQL expression or aggregate function. For some chart types (like Mixed Timeseries), there's also a "Show query identifiers" checkbox—if you turn this off, legend labels will omit SQL expressions and show only the metric display name and dimension values. See code reference, Mixed Timeseries PR.

So, to get the legend labels you expect, add custom labels to your metrics or check for relevant chart options in the control panel.

To reply, just mention @dosu.

---

^{How did I do? Good | Irrelevant | Incorrect | Verbose | Hallucination | Report 🐛 | Other}    

[MidnightOakMapleNote]

@dosu thank you for your reply , how to set a custom "label" or "verbose name" for each metric in my chart configuration

[dosubot[bot]]

You can set a custom label (verbose name) for each metric in your chart configuration from the Explore view:

1. In the chart configuration panel, find the "Metrics" section.
2. For ad-hoc metrics (those you define directly in the chart, not saved in the dataset), click on the metric you want to edit. This opens a popover or dialog.
3. In the popover, look for the "Label" or "Custom Label" field. Enter your desired display name here.
4. Press Enter or click "Save" to confirm. The new label will now appear in the legend and other chart elements instead of the SQL expression or default aggregate name.

If you want to remove the custom label and revert to the default, clear the label field and save again.

This workflow is consistent across chart types, including eCharts. The label you set here will be used in the chart legend and tooltips, making your visualizations much clearer and more user-friendly.

For saved metrics (defined in the dataset), you can set the "verbose name" when creating or editing the metric in the dataset editor. This name will also be used in legends and labels.

Let me know if you need a screenshot or more details for a specific chart type!

To reply, just mention @dosu.

---

^{How did I do? Good | Irrelevant | Incorrect | Verbose | Hallucination | Report 🐛 | Other}    

[MidnightOakMapleNote]

Hello, thank you for your reply. There is such a scenario: multiple charts are configured for cross-filtering, and one of the charts is set up with one dimension and two metrics. If custom labels are added to the metrics, the corresponding filtering SQL statement becomes "WHERE customer_name IN (custom labels)". If no custom metrics are configured, the filtering SQL becomes a statement like "WHERE customer_name IN (customer_name IN ('COUNT_DISTINCT(warehouse_type)'))". What should be done in this case?
@dosu

[dosubot[bot]]

This is a known limitation in Superset's cross-filtering logic for charts with multiple metrics. When you use more than one metric, the legend labels are built by combining the metric label (or SQL expression) with the dimension value. During cross-filtering, Superset tries to map the clicked legend label back to the dimension value using a labelMap, but in multi-metric scenarios, this mapping includes the metric label as part of the filter value. As a result, the generated SQL filter ends up using metric labels or SQL expressions instead of just the dimension values, which leads to the confusing SQL you observed (code reference).

Currently, there isn't a built-in workaround to force Superset to filter only by dimension values in this scenario. The safest options are to use only one metric per chart when cross-filtering is needed, or to avoid relying on cross-filtering for charts with multiple metrics. This limitation affects all eCharts-based visualizations with multi-metric groupby setups.

If you need cross-filtering with multiple metrics, you may want to monitor future Superset releases for improvements in this area, as the frontend and backend would need changes to better separate metric and dimension values in legend and filter logic.

To reply, just mention @dosu.

---

^{How did I do? Good | Irrelevant | Incorrect | Verbose | Hallucination | Report 🐛 | Other}