Replace dataset refers to the ability to quickly switch the dataset for a dashboard after it has been created, using the new dataset to update the dashboard. In addition, after using a dashboard template, you can retain the template's elegant style while replacing the original dashboard data with your own business data.
Common application scenarios

• When the underlying dataset changes, you can replace the underlying data while keeping the charts in the dashboard unchanged.
• Dashboards are designed to be general-purpose, with the intention of replicating them horizontally across multiple business lines. You can copy dashboards and replace the datasets in the copied dashboards, quickly completing the workflow from dataset creation, to chart building, to dashboard assembly.

Current methods supported for replacing datasets in charts

• Project Center Chart Management page: Allows batch replacement of datasets for charts under the project.
• Dashboard page: Allows batch replacement of datasets for charts on the dashboard (currently, replacing the dataset for a single chart on the dashboard is not supported).
• Visualization Query page: Allows replacement of the dataset for a single saved chart.

Step 1: Replacement entry point #

Path 1: In the Project Center - Chart Management module, after filtering for the dataset to be replaced, click the chart whose dataset you want to replace, then click the [Replace Dataset] icon on the right.

Note: Only project administrators, chart owners, and dashboard administrators where the chart resides can perform dataset replacement.

Users can also select multiple chart names to batch replace datasets. Caution: Only batch replacement for a single dataset is supported.

Path 2: In the dashboard editing interface, click "Replace Dataset" in the upper right corner to batch replace datasets for charts on the dashboard.

Step 2: Replacement configuration #

Complete the dataset replacement configuration. After selecting the target dataset for replacement, perform model field mapping, which includes data source field mapping (required) and dataset field mapping (optional). For a detailed understanding of data source and dataset fields, refer to Fields and Field Expressions FAQs. For the specific mapping results, see the detailed feature introduction below.
Note: By default, the system automatically matches fields by field name. If there is a data type mismatch, the model will be highlighted in red. Please ensure the types are consistent before proceeding to the next step.

Step 3: Complete the replacement #

After confirming the mapping information, click "Next" to open the field confirmation page. If the original dataset fields used by the chart do not have fields with the same name in the target dataset and no dataset field mapping is configured, the corresponding fields will be automatically added to the new dataset.

3.1 Dataset field mapping feature #

When configuring chart dataset replacement, since each data source field requires a mapping relationship, if the current chart contains a large number of fields and dataset replacement operations are frequently performed,
manually configuring field mapping for each field during every dataset replacement can be very cumbersome and incurs a high operational cost.
To address this, a field mapping configuration is provided, allowing you to pre-configure mapping relationships between datasets. When performing chart dataset replacement operations on these datasets with mapping configured, the mapping relationships will be automatically populated according to the configuration, reducing operational costs.

For detailed operating instructions, see Field Configuration.

3.2 Supported dataset types for replacement #

• Supported:
  • Standard datasets (including regular extracted and direct connection datasets), and mirror datasets
• Not supported:
  • Tag datasets, group datasets
  • Charts created based on visual JOIN

3.3 Replacement results #

Data source field mapping mainly refers to replacing the fields on the chart and the data source fields of the old dataset referenced in expressions with the data source fields of the new dataset.

• If a field from the old dataset has a field with the same name in the new dataset, it will be directly replaced with the field from the new dataset, and the calculation logic of the new dataset field will be used directly.
• If a field from the old dataset does not have a field with the same name in the new dataset, the corresponding field will be automatically added to the new dataset during replacement, and the logic of the new field will inherit the logic of the old dataset field.

Dataset field mapping allows the fields in the current chart to be completely replaced with a field from the new dataset.

• This applies to scenarios where the new and old datasets have fields with the same name, but to replace field A from the old dataset with field B from the new dataset.

3.3.1 Data source field mapping #

Data source field mapping mainly refers to replacing the fields on the chart and the data source fields of the old dataset referenced in expressions with the data source fields of the new dataset.

Scenario 1: No expression nesting

First is a relatively simple scenario, where the expressions in the chart are simple and do not reference other calculated fields. Here, we use the following chart as an example:

• There are two fields in the chart, "sale_province" and "price".

• The expressions for these two fields are as follows:

• After replacing the dataset for this chart, the mapping relationship is as shown in the figure below.

• After the replacement is completed, the corresponding data source fields of the target dataset will replace the original dataset's data source fields in the expressions, as shown in the figure below.

Scenario 2: Expression nesting exists

This scenario is also commonly encountered in actual business, where the expressions of fields in the chart reference other fields. Let us use the following query as an example:

• There is a calculated field "GMV" in the chart, which involves the dataset fields "order_count" and "sale_price".

• After replacing the dataset for the chart, the mapping relationship is shown in the figure below. Although there are only two fields in the chart, since the fully expanded underlying expression of [order_count]*[sale_price] references other data source fields, all these data source fields must be replaced as well.

• After replacement, all data source fields in the expression of the dataset field, after fully expanding it, will be replaced with the target dataset fields, as shown in the figure below. Although the expression for the "GMV" field appears unchanged, its referenced underlying fields "order_count" and "sale_price" will be replaced.

Scenario 3: Matching data source fields to dataset fields

When replacing the dataset for a chart, in addition to replacing between data source fields, you can also choose to replace the data source fields of the current dataset in the field expression with dataset fields of the target dataset.
This applies to scenarios where the calculation logic for fields differs between the old and new datasets.

• Take the following figure as an example. In the chart, the expression for the "sales" field is the data source field 'sales'.

• When replacing, we choose to replace the data source field sales of the current dataset with the dataset field "sale_price" of the target dataset.

• After replacement, sales in the expression of the "sales" field will be replaced with the dataset field "price-calculation".

3.3.2 Dataset field matching #

Data source field matching mainly replaces the data source fields referenced in field expressions.
Dataset field matching, on the other hand, allows you to directly and completely replace a field in the current chart with a field from the new dataset.
Take the following chart as an example. The chart contains two dataset fields, named "sale_province" and "price":

• At this point, we replace the dataset and configure dataset field matching, replacing "price" with "sales" from the new dataset, while leaving "sale_province" unchanged.

• In this case, even though the new dataset does not have "price" and "province" fields, you only need to add a "province" field.

• After the replacement is completed, the original "price" field in the chart will be directly replaced with the "sales" field from the new dataset. The calculation logic for this field will also use the logic from the new dataset, and the "price" field will not be added again to the new dataset.

3.4 Other impacts #

3.4.1 Dataset parameters #

• There is no impact on project-level "personal parameters".
• For dataset-level "public parameters", if they were previously applied to the original dataset, after replacement, the scope of application will include the new dataset that is used as a replacement.

3.4.2 Hierarchy and category #

The hierarchy and category of the original dataset will not be copied to the new dataset, and the hierarchy and category of the new dataset will remain unchanged.

3.4.3 Permissions #

After replacement, whether the permissions of the original dataset are appended and inherited to the new dataset depends on the settings.

3.4.4 Dashboard #

For charts that have had their dataset replaced, or dashboards containing charts with replaced datasets, in principle, all components and configurations will remain unchanged. This includes public filter components, linking and navigation configurations, chart colors, legends, and other configurations.

4.1 Field data type mismatch #

You must ensure that the model field data types of the datasets before and after chart replacement are consistent.

• As shown in the screenshot below, if "order_date" is a date-type field and "GMV" is a numeric-type field, a field data type mismatch error will occur.
  

When the above error message appears, users can take the following actions to adjust and optimize:

• Confirm whether the field type of the target dataset to be used as a replacement meets expectations. If modification is needed, modify the field type in the upstream table of the dataset.
• Caution: The field to be modified here is the model field of the dataset, not the dataset field itself. The model field refers to the field defined in the dataset's data model, while the dataset field refers to the field as it appears in the dataset.
  • If the dataset uses a Lark table as the data source, modify the data type of the corresponding column in the Lark table.
  • If the dataset uses Hive as the data source, modify the data type of the corresponding field in the Hive table.
  • If the dataset is a custom SQL model dataset, conversion can be performed in SQL using the CAST function.

4.2 Field information retrieval failed #

The following reasons may cause field information retrieval to fail:

• The chart in the current dashboard is not only placed on this dashboard, but also on other dashboards. The user replacing it does not have edit permissions for those dashboards.
• There are too many charts on the current dashboard.