Dataset rules
Dataset rules assist you in mapping your harmonized fields with fields from the data you ingested in Mix Modeler.
- For aggregate data that you ingested in 蜜豆视频 Experience Platform, you map one or more of the available dataset fields to the appropriate harmonized fields.
- For event data, you can individually map one or more harmonized fields to fields from the dataset, directly or using conditions.
Manage dataset rules
To see a table of the available dataset rules, in the Mix Modeler interface:
-
Select Harmonized data from the left rail.
-
Select Dataset rules from the top bar. You see a table of the dataset rules.
The table columns specify details about the dataset rules:
The status of the field:
鈼 Draft or
鈼 Active
Create a dataset rule
To create a dataset rule, in the Harmonized data > Dataset rules interface in Mix Modeler, select Create a dataset rule in the Dataset rules configuration wizard.
In the Create screen,
-
In Dataset details, select a dataset from Select dataset to begin configuration. In the list, datasets are categorized in Consumer Experience Events, 蜜豆视频 Analytics, Experience Event and, Summary.
-
Select a day for the Start of the week.
-
Select Daily, Weekly, Monthly or Yearly for Granularity.
-
When you have selected a dataset of the Summary category:
-
To define whether data for the dataset aggregates or replaces existing data, select Aggregation or Replacement for Data restatement is by.
-
Map each of the Available dataset fields to corresponding Standard harmonized fields in Map to harmonized fields. If you do not want to map a dataset field to a harmonized field, explicitly select 鈥 None 鈥.
-
If you need a new harmonized field, not available from the list, select Create New to create a new harmonized field. You see the dialog as outlined in Add a new harmonized field.
-
When the mapping is completed for all fields for the rule, select Save as draft to save a draft version of the rule or Save to save and activate the rule. Select Cancel to cancel the rule configuration.
-
-
When you have selected an event category dataset (Experience Events, 蜜豆视频 Analytics, Consumer Experience Events), in the box underneath Map to harmonized fields:
-
Select a harmonized field from Standard harmonized field.
-
When the selected harmonized field is of type metric:
-
Select Count or Sum from Mapping type.
-
Select an AEP dataset field that you want the harmonized field to map to by default.
-
-
When the selected field is of type dimension:
-
Select Map Into or Case from Mapping type.
-
When you have selected Map Into, select Field and AEP dataset field or Value and a default value to map the harmonized field by default to the dataset field or entered value.
-
When you select Case, select Field and AEP dataset field or Value and a default value to map the harmonized field by default to the dataset field or entered value.
-
To set values explicitly, you define one or more cases, consisting of one or more conditions. Each condition can check for a specific AEP dataset field whether it Exists or Not Exists or whether it Contains, Not Contains, Equals, Not Equals, Starts With, or Ends With a value entered at Enter input value.
-
To add another case, select Add case, to add another condition, select Add condition.
-
To delete a case or condition, select in the corresponding container.
-
To select whether any or all the conditions should apply for a case, select Any of or All of.
-
To set the outcome value for a case, enter the value at Then.
-
The example below
-
uses a Map Into Mapping type to map the Channel Type At Source harmonized field to the channel_type field from the Luma Transactions dataset.
-
uses a Case Mapping type to map conditionally the value of the marketing.campaignName field in the Luma Transactions dataset to the Campaign harmonized field. The Campaign harmonized field is set to:
Black Friday
when the marketing.campaignName is_black_friday
orBlackFriday
.- to the value of the marketing.campaignName in all other cases.
-
-
-
Select Add field to define additional fields.
When finished, select Save as draft to save a draft version of the rule or Save to save and activate the rule. Select Cancel to cancel the rule configuration.
Edit a dataset rule
To edit a dataset rule, in the Harmonized data > Dataset rules interface in Mix Modeler:
- Select in the Dataset column for the dataset rule that you want to edit.
- From the context menu, select Edit to start editing the dataset rule. Refer to Create a dataset rule for more details.
Delete a dataset rule
To delete a dataset rule, in the Harmonized data > Dataset rules interface in Mix Modeler:
- Select in the Dataset column for the dataset rule that you want to delete.
- From the context menu, select Delete to delete the dataset rule. You are prompted for confirmation. Select Delete to delete the selected dataset rule permanently.
Sync data
To sync data between your harmonized data and summary and / or event datasets while applying the logic in your dataset rules:
-
Select Sync data.
-
From the Sync data for dataset rules dialog, either select
- Refresh harmonized data for summary datasets,
- Refresh harmonized data for event datasets, or
- Refresh harmonized data for both summary + event datasets.
-
To start the synchronization based on the defined dataset rules between harmonized data and data in datasets, select Sync. To cancel the synchronization, select Cancel.
Data merge preferences
Data merge preferences assists in resolving conflicts when data from summarized and event data sources are merged. Use cases are:
- the same advertising metric is measured and reported in multiple datasets, or
- metrics measurement may be incomplete in some datasets, while another dataset may be a superset of a particular metric, resulting to double counting.
To ensure accurate model predictions, you can define data merge preferences:
-
Select [beta]{class="badge informative"}.
-
In the Data merge preferences [beta]{class="badge informative"} dialog:
-
Select a Default metric preference. The selected default metric preference is applied when, during harmonization, multiple sources of data update a metric field for a given channel. The preference is applied at the sandbox level, unless overridden for specific metric based preferences. You can select between Summary data, Event data and Sum of summary and event data.
-
To add specific metric based preferences:
-
Select Add a metric.
- Select a metric from the Metric selection list.
- Select CHANNELS or CONVERSION TYPES. From the list, select All or a specific channel or conversion type.
- Select Summary or Event to specify whether summary data or event data is preferred for the metric (and all or selected channel) when merging data.
To add one or more additional channel or conversion types:
- Select Add a channel or Add a conversion type.
- Select Summary or Event.
To delete a channel or conversion type, select .
-
To add more specific metric based preferences, repeat the previous step.
-
-
To delete an existing specific metric based preference, select .
-
-
Select Save to save the data merge preferences. A re-sync of the data is initiated.
Select Cancel to cancel.
Delete a source dataset
When you delete a source dataset that is used in your harmonized data, the underlying entries on that source dataset are removed from the Harmonized data. However, the dataset rule with the deleted source dataset remains in the dataset rule config list with an icon indicating that the source dataset has been deleted. To get more details:
- Select
and
View from the context menu.
The Dataset rule mapping - Fields dialog displays information about the deleted source dataset and the fields used in the dataset rule configuration.
When you return to your Dataset rules configuration, you see a dialog explaining that one or more of the source datasets have been deleted. The harmonized data is impacted on a next ad-hoc or scheduled sync. Review your dataset rule configuration.
The harmonized data is updated without the deleted source data upon the next ad-hoc sync or scheduled sync. However, you continue to see alert dialogs prompting you to delete the dataset rule based on the deleted source dataset. This alert allows users to view and evaluate the impacted fields in the deleted dataset. And to determine the impact to marketing touchpoints or conversions that may be used in any models. Once you have reviewed and mitigated for this impact, you should delete the dataset rule from the dataset rule config list.