🗂️ Launched: Custom Dataset Views

:card_index_dividers: Upcoming: Dataset Views

Note: This is an early feature launch announcement, meaning the feature is still in development. We make this announcement early to gather feedback and make sure what we build actually helps solve our users’ use cases.

To address the following feature requests:

We are working on an upcoming feature Dataset View, an enhancement to Holistics Datasets. This feature aims to let dataset creators design the ideal self-service experience for their end users by allowing them to customize the display of fields and models within datasets.

How it works: By default, datasets in Holisitics display all data models and fields in the order listed in the AML files. Now with Dataset View, creators can:

  • Select fields: choose which fields to display in the dataset UI
  • Re-order fields and models: define a customized order for fields and models within the dataset
  • Group models or fields: put fields or models inside groups

Creating Dataset view in AML

Behind the scenes, Dataset views utilizes a view parameter in the dataset definition, which allows dataset creators to customize the display of information. Below is how a dataset view can be defined in the dataset file:

Dataset ecommerce {
  models: [customers, products, orders, dim_dates]
  relationships: [
	relationship(orders.customer_id > customers.id, true),
	relationship(orders.product_ids > products.id, true),
	relationship(orders.order_date > dim_dates.id, true)

  view {
	model products { } // Display all fields from model products

	model customers {
		// Display selected fields from model customers
		field id
		group Customer_Name {
			field first_name
			field last_name
			field full_name
		field email
		field address

	group Order_Master {
		model orders {
			field order_value
			field id
			field order_status
			field order_discount

		model dim_dates {
			field year
			field quarter
			field week_number
			field week

In this example, the dataset ecommerce consists of four data models customers, products, orders, and dim_dates. It is customized using the view parameter to:

  • Display all fields from products
  • Only display selected fields from customers
  • Create a group called Order_Master containing selected fields from models orders and dim_dates

By defining the view in this way, dataset creators have fine-grained control over the organization and presentation of fields and models.

Using Dataset View in Exploration

Dataset views simplify data organization while maintaining the integrity of data models and relationships within the semantic layer. This makes it easier for business users to get started with curated datasets.


Release Timeline

This feature is currently under development and is expected to be released by mid Q3 2023.

Please comment below or reach out if you’re interested in this feature. Also, please share some of your use cases that would benefit from Dataset View.


:card_index_dividers: Launched: Dataset view in Open Beta

Hi there,

We are excited to announce that Dataset View is now in Open Beta - it’s available to everyone using Holistics 4.0 version right away (no sign up needed)!

This version of Dataset view allows dataset creators to:

  • Select fields: choose which fields to display in the dataset UI
  • Re-order fields and models: define a customized order for fields and models within the dataset

For more information, please watch our demo video:

To create a dataset view, go to .dataset.aml file and follow this syntax:

Dataset raw_ecommerce {
  models: [users, orders]
  relationships: [relationship(orders.user_id > users.id, true)]

  view {
    model orders // Display all fields from model orders. Put model orders first
    model users { // Display selected fields from model users
      field id
      field first_name
      field last_name
      field full_name
      field sign_up_at
      field acquisition_cohort
      field number_of_orders

Open Beta

This feature is available in Open Beta and we’ll be back with more updates for this exciting new feature.

:point_down: Feel free to share your feedback with us down in the comments.


:card_index_dividers: Launched: Custom Dataset Views

Previously, we shared with you an initial glimpse into our development process for the Dataset View feature in this early announcement. After two months of development, we’re glad to announce the official launch of Custom Dataset Views.

Here’s what we arrived at:


When designing new datasets, it can be quite challenging to strike the right balance between “comprehensiveness” and “usability”.

  • Comprehensiveness: It’s important that the dataset involves a wide range of relevant information to provide a comprehensive view.
  • Usability: The dataset should also be user-friendly, allowing users to quickly find the information they need. A usable dataset should be well-organized and easy to navigate.

To address this challenge, we introduce a feature called Custom Dataset Views. This feature allows data analysts to curate the display of fields and models within datasets. With custom dataset view, analysts can:

  • Reorder fields and models to highlight frequently used models and fields by putting them on top.
  • Hide fields and models that are not relevant to end users.
  • [New] Group relevant fields and models to provide better context.

How it works

Creating views in dataset defnition

By default, in dataset UI:

  • Fields are grouped by dimensions and measures, then displayed by the order listed in model files.
  • Models are displayed by the order listed in dataset files.

To create a custom dataset view, data analysts can define a view parameter in the dataset definition.

Example syntax
Dataset ecommerce {
  label: 'Ecommerce'
  models: [users, products, orders, order_items, dim_dates, categories]
  relationships: [
    relationship(orders.user_id > users.id, true),
    relationship(order_items.product_id > products.id, true),
    relationship(orders.created_date > dim_dates.date_key, true)
    relationship(products.category_id > categories.id, true)

  view {
    model orders {
      // field id
      field status
      field created_at
      field user_id
      field discount
      field delivery_attempts

      group order_status_tracking {
        field created_date
        field total_orders_count
        field delivered_orders_count
        field cancelled_orders_count
        field refunded_orders_count
    model order_items
    model users

    group product_info {
      model products
      model categories
    // model dim_dates

Using Custom Dataset Views in Exploration

When a dataset has custom views, the view is displayed by default. This makes it easy for end users to get started with curated datasets. You still have the option to switch to the original field list if you prefer.

For more details information, check out our documentation: Custom Dataset View | Holistics Docs

Notes: Custom Dataset Views is only available for customers on Standard plans and above for Holistics version 4.0. If you have any feedback or suggestions for future updates, please don’t hesitate to let us know. :point_down:


Great feature! Is it possible to collapse the views by default? I’m trying to make the ‘start’ view as simplified as possible and prevent too many dimensions to appear when user opens the page.

1 Like

Hi @Engineer,

Great to hear that you’re like this feature.

Thank you so much for bringing this to our attention. We are currently refining the collapse behavior so that only the first model or group will be expanded when initially viewing a dataset (similar to the original dataset view). This change is anticipated to be available in the coming weeks. Rest assured, we’ll keep you updated.

Will this solution meet your needs? If not, could you please provide further details of your ideal solution? Your feedback is invaluable to us :love_you_gesture:

Hi Antti and everyone,

We are delighted to share that we have just rolled out an update to the Custom Dataset View. Now, only the first model is expanded, while others are collapsed by default. This adjustment is designed to provide Explorers with a cleaner view of the dataset, making it easier to get started.


We hope you enjoy this improvement.

Cheers :beers:

1 Like