In , a target is the set of columns, their order, and their formats to which you are attempting to wrangle your dataset. This target can be defined through imported or created datasets and must be assigned to an existing recipe. After it is assigned to a recipe, a target appears in the Transformer page to assist in your wrangling efforts. You can also apply changes to selected columns based on the target.
In general, a target consists of the set of information required to define the expected data in a dataset. Often referred to as a "schema," this target schema information can include:
A dataset associated with a target is expected to conform to the requirements of the schema. Where there are differences between target schema and dataset schema, a validation indicator (or schema tag) is displayed.
In , a target is created from the information in a dataset and can be applied to a recipe in a flow. When you are working with the flow, the target information is available for your wrangling activities, so that you can match up columns in your dataset (source) with their corresponding columns in the target. As you make changes in your recipe through the Transformer page, the target schema is available as a reference to see if your latest changes get you closer to matching the dataset to the target.
Targets are applied only after initial type inferencing has been applied to the loaded dataset.
Tip: As needed, you can disable initial type inferencing when data is imported into the product. See Import Data Page.
settypetransform to any selected column. No pattern matching or standardization is applied. For more information, see Overview of Pattern Matching.
Changes to the underlying objects of a target schema are not reflected in the schema. A target schema is a snapshot of the object at the time of its creation. To update, delete the target and create a new one.
Tip: If your target schema source is a recipe, then you can modify the recipe as needed and use it as your target again.
The schema used to define a target can be imported and assigned from any of the following objects, including:
An imported dataset
Ideally, the source of the target schema should come from the publishing target. If you are publishing to a pre-existing target, you can create do one of the following:
You can create a target through one of the following mechanisms:
For more information, see Create Target.
After a target has been attached to a recipe, the target schema appears in a toolbar above the data grid along with a preview of the data. You can then make modifications to the data so that each column matches the definition for the corresponding column in the schema. See Data Grid Panel.
Through the data grid and the Column Browser, you can perform operations on selected columns in your dataset to align them with the target schema. For more information, see Column Browser Panel.
NOTE: You can run a job even if there are differences between the schema and your dataset. In , no error checking is performed between schema and data prior to job execution. If you are publishing to a target that has a predefined schema, a publication error may be generated.
You can experiment with fuzzy matching thresholds to ensure that matches are occurring properly. This parameter applies a specific threshold value when two values are compared for matching. Lower values increase the probability of a match.
Adjust the value between 0.00 and 0.99 for the following parameter:
If you prefer to disable this feature, please complete the following steps.
NOTE: If you are experiencing performance issues with target matching, you can first try to disable fuzzy matching, which can be resource-intensive.
Tip: If there is no schema associated with a recipe, then the target schema matching features are not displayed.
Set the following parameters to
"feature.targetMatching.enabled" : false, "feature.targetMatching.fuzzyMatchingEnabled" : false,