Tutorial: Create a decomposition tree with a Power BI sample

Article
03/08/2023
3 minutes to read

APPLIES TO: Power BI Desktop Power BI service

In this tutorial, you start with a built-in Power BI sample dataset and create a report with a decomposition tree, an interactive visual for ad hoc exploration and conducting root cause analysis. It's also an artificial intelligence (AI) visualization, so you can ask it to find the next category, or dimension, to drill down into based on certain criteria.

Screenshot showing the completed decomposition tree. This tutorial shows you how to:

Sign up for a Power BI license, if you don't have one.

Import the Retail Analysis sample and add it to the Power BI service.

Create a decomposition tree.

Do root cause analysis on your data in the decomp tree in Edit mode.

Save the report and continue root cause analysis in reading view.

If you want to familiarize yourself with the built-in sample in this tutorial and its scenario, see Retail Analysis sample for Power BI: Take a tour before you begin.

Prerequisites

To download a sample in the Power BI service, you can sign up for a free or trial license.

Import the sample in the Power BI service

Open the Power BI service (app.powerbi.com), sign in, and open the workspace where you want to save the sample.

If you don't have a Power BI Pro or Premium Per User (PPU) license, you can save the sample to your My Workspace.
Select Get data at the bottom of the nav pane.

If you don't see Get Data, expand the nav pane by selecting the following icon at the top of the pane.
On the Get Data page that appears, select Samples.
Select Retail Analysis Sample > Connect.

Create a report from the dataset

On the Datasets + dataflows tab, you have several options for exploring your dataset. In this tutorial, you're going to explore the dataset by creating your own report from scratch.

Screenshot showing the Retail Analysis Sample dataset entry.

Select More options (...) > Create report.

The dataset opens in report editing mode.

Create a decomp tree

Let's add a decomposition tree, or decomp tree, to our report for ad hoc analysis.

Open Power BI Desktop and load the Retail Analysis Sample. You can get this sample from Download original sample Power BI files.
Select the Report icon to open the Reports view. In the Visualizations pane, select the Decomposition tree icon.
Drag the edge so it fills most of the page. Decomposition trees can get wide.
Decomp trees analyze one value by many categories, or dimensions. Expand Sales > This Year Sales and select Value. Power BI adds Value to the Analyze box.
Next, select dimension fields and add them to the Explain by box. Add as many as you want, in any order. You can use them or not, in any order, in the decomp tree. We added:
- District Manager
- Category
- FamilyName
- Segment
- Chain
- PostalCode
- Store Type
- Territory

Analyze in the decomp tree

Now comes the ad hoc analysis part.

Select the plus sign (+) next to This Year Sales and select High value. That means Power BI will use artificial intelligence to analyze all the different categories in the Explain by box, and pick the one to drill into to get the highest value of the measure being analyzed.

Power BI selects Store type.
Keep selecting High value until you have a decomp tree that looks like this one.

Or select other values yourself, and see what you end up with.
You can delete levels by selecting the X in the heading.
Save your report. Now anyone who views your report can interact with the decomp tree, starting from the first This Year Sales and choosing their own path to follow.