Decomposition Tree - AI Visuals in Power BI

Angelica Choo Quan, a trainer at Pragmatic Works, continues her Power BI AI Visuals series by exploring the Decomposition Tree in Power BI Desktop. This powerful visual tool simplifies data analysis by allowing users to break down data across multiple dimensions dynamically. Angelica provides a detailed walkthrough of how to create and customize a decomposition tree to analyze business data effectively.

What is the Decomposition Tree?

The decomposition tree visual in Power BI allows users to visualize data across multiple dimensions, automatically aggregating results and enabling users to drill into data for deeper insights. It helps explore factors contributing to a specific metric, such as profit or sales, by splitting data into layers based on various dimensions.

Setting Up the Decomposition Tree

Angelica begins by explaining the initial setup of a decomposition tree visual:

1. Open the Power BI Desktop and locate the decomposition tree visual in the visualization pane.
2. In the Analyze field, input the metric to be measured (e.g., profit).
3. In the Explain By section, add the dimensions to break down the data (e.g., Country, Region, Product Model Name, and Color).

This basic setup allows the visual to begin displaying how different dimensions contribute to the overall metric.

Drilling into Data with the Decomposition Tree

The decomposition tree provides a flexible way to explore data through customizable nodes. Users can drill into data manually or use Power BI’s AI features for automatic data breakdown:

• Manual Exploration: Click the plus sign next to a node and choose specific dimensions to drill into (e.g., analyzing profit by country or product model).
• AI-Powered Exploration: Choose "High Value" or "Low Value" to allow Power BI to highlight the most significant factors contributing to the selected metric automatically.

Angelica demonstrates how to break down profit by country and further explore regions and product models using both manual and AI-driven options.

AI Splits: High Value vs. Low Value

Angelica emphasizes the importance of AI splits, which are unique to the decomposition tree:

• High Value Split: Automatically identifies dimensions contributing the most to the selected metric.
• Low Value Split: Highlights dimensions with the least contribution to the metric.

These AI splits are visually distinct, represented by a dashed line, compared to solid lines for manually chosen dimensions.

Locking and Customizing the Decomposition Tree

The decomposition tree can be locked to restrict user interaction. This is particularly useful for sharing reports with restricted exploration permissions:

• Locking prevents viewers from modifying the dimensions used in the tree.
• Users can also unlock dimensions for further exploration if desired.

Angelica showcases how to lock dimensions and adjust the visual to control the level of interaction available for end-users.

Interactive Features and Filtering

One of the decomposition tree's strengths is its ability to respond to filters and slicers applied to the report page. Angelica demonstrates how selecting specific years or months adjusts the data displayed in the tree dynamically.

Drill-Through Setup in Power BI

Angelica introduces the concept of creating a drill-through page to provide detailed insights when clicking a data point within the decomposition tree:

1. Create a new report page and name it appropriately (e.g., "Detail Page").
2. Drag fields matching those used in the decomposition tree onto the drill-through page.
3. Enable drill-through functionality by adding the key metric (profit) as a drill-through field.
4. Right-click a node in the decomposition tree and select "Drill Through" to navigate to the detailed view.

This feature allows users to explore a specific subset of data in greater depth.

Decomposition Tree Limitations

Angelica highlights important limitations of the decomposition tree visual:

• Maximum Levels: The visual supports up to 50 levels of data breakdown.
• Data Points: A maximum of 5,000 data points can be visualized at once.
• Compatibility: AI splits are not supported in on-premises Analysis Services, Power BI Report Server, or "Publish to Web" reports.

Conclusion

Angelica concludes by emphasizing the decomposition tree's value in data analysis and decision-making. This AI-powered visual helps users identify patterns, outliers, and trends within their data effortlessly. She encourages viewers to experiment with both manual and AI splits for a deeper understanding of their datasets.

Don't forget to check out the Pragmatic Works' on-demand learning platform for more insightful content and training sessions on Power BI and other Microsoft applications. Be sure to subscribe to the Pragmatic Works YouTube channel to stay up-to-date on the latest tips and tricks.