Column actor dataset
In Analytics, an actor is any entity in your business commonly used in analysis, such as a user, account, subscription, product, and so on.
A dataset represents an actor – typically, this maps to a dimension table in the data warehouse. However, in some cases, there is no such dimension table in the warehouse; rather, there is only an identifier column in an event table such as
(Events.user_id)In this case, you can create a column actor dataset. It does not have any direct mapping to a warehouse table; it is defined solely by its relationship to the identifiers in other tables. However, you can still use column actors in all the ways you can use a regular actor dataset.
Create a column actor dataset
From the datasets page
- Go to Data > Datasets > + New Dataset > Column Actor Dataset.
- Name your dataset, add a description, and click Save. Then, configure the dataset.
Within explorations
You can also create column actor datasets inline – that is, within the exploration templates. Wherever there is a requirement for actor selection within any exploration template, the inline column actor creation option is available within the terminal selection. To create a column actor dataset within an exploration,
- Click + on the navigation panel and choose an exploration template.
- Select a measure and the actor. Open the drop-down list and click the Create Column Actor from an ID column option in the terminal selection.
- Select the related dataset and columns, choose the destination folder, and click Save.
Note
Suppose you have already selected the events for the Funnel stages. In that case, when you click the Create Column Actor from an ID column option, you preselect the related dataset – that is, you preselect the event dataset of the events selected in the stages. Otherwise, the app dataset would be preselected, if that exists, and if none of these are present, then it would be empty, and you can choose a dataset of your choice.
Dataset configuration
Related datasets
The Related datasets tab lets you define logical relationships between datasets, similar to how tables relate in a data warehouse. You can combine event and attribute data for richer analysis without understanding the underlying data structure. Analytics uses these relationships to determine which datasets to join in an exploration and only shows relevant options in the UI. For example, using common user ID columns, you can form a relationship between an events dataset and a users dataset.
- Click + Add Related Dataset.
- From the Dataset drop-down list, select the dataset you want to link to the current source dataset.
- Select the relevant Cardinality for the relationship:
- many to one – Match many rows in the current dataset to one row in the related dataset.
- one to many – Match one row in the current dataset to multiple rows in the related dataset based on the selected columns.
- one to one – Establish a relationship by directly matching a selected column in a single row of datasets.
- Click + and use the drop-down lists to select the columns that must match to establish a successful relationship.
- Click Save.
Health check
The Check health option in the Related Datasets tab evaluates a defined relationship between two datasets. It identifies how many distinct values on the many side of the relationship do not have a corresponding match on the one side. This helps uncover broken links or orphaned records that could impact data integrity.
For datasets identified as event streams, the check automatically filters data to the last 60 days. To ensure predictable performance and cost efficiency, the scan is limited to a maximum of 10 million rows.
There are three different health statuses possible, and they are as follows:
- Healthy – Indicates <10% of unmatched records, suggesting the relationship is mostly intact. For example, when the status is healthy, it displays the following message: Join null rate: 8% (80 nulls out of 1000 rows).
- Warning – Indicates 10-20% of unmatched records, which could impact data integrity. For example, when the status is a warning, it displays the following message: Join null rate: 15% (150 nulls out of 1000 rows).
- Unhealthy – Indicates >20% of unmatched records, signaling a critical data integrity issue. For example, when the status is unhealthy, it displays the following message: Join null rate: 25% (250 nulls out of 1000 rows).
Note
The health check option is not available for one-to-one cardinality.
Cohorts and derived columns
In this section, you can define new derived columns and cohorts of rows that represent logical groupings within the respective tabs. This is an optional step. Learn more about cohorts and derived columns.
You can view and access your dataset on the Datasets page when the configuration is complete.
Edit your dataset
- Go to Data > Datasets.
- Click the dataset you want to edit on the Datasets page.
- Make necessary changes and click Save.
Duplicate your dataset
- Go to Data > Datasets.
- Click the dataset you want to duplicate on the Datasets page.
- Click More (⋮) > Save As. Give your copy a name and a description, choose a destination folder, and click OK.
Delete your dataset
- Go to Data > Datasets.
- Click the dataset you want to delete on the Datasets page.
- Click More (⋮) > Delete. Once again, click Delete in the confirmation box.
