AI-Powered Visuals: Key Influencers & Decomposition Tree

Microsoft has invested billions of dollars integrating automated machine learning directly into Power BI visuals. Rather than requiring users to manually formulate hypotheses, AI Visuals scan underlying data patterns to explain root causes and decompose metrics automatically.

1. The Key Influencers Visual: Automated ML Regression

The Key Influencers visual analyzes an outcome metric (e.g., Customer Churn, Late Deliveries, High Employee Satisfaction) and uses built-in logistic or linear regression to tell you:

• "What factors drive a customer to Churn?"
• "What factors increase sales margin?"

+-----------------------------------------------------------------------------+
| KEY INFLUENCERS: What influences Customer Churn to be "Yes"?               |
|                                                                             |
| When...                                    The likelihood of Churn is...    |
| Contract is Month-to-month  -------------> 4.2x More Likely                 |
| Internet Service is Fiber optic ---------> 2.8x More Likely                 |
| Payment Method is Electronic check ------> 2.1x More Likely                 |
| Tenure in Months is < 6 -----------------> 1.9x More Likely                 |
+-----------------------------------------------------------------------------+

How to Configure Key Influencers:

1. 1
   Insert the Key Influencers visual on the canvas.
2. 2
   Drag the target outcome into Analyze (e.g., Dim_Customer[Churn]).
3. 3
   Drag candidate explanatory drivers into Explain by (e.g., Contract, TenureMonths, MonthlyCharges, TechSupport).
4. 4
   Power BI automatically runs regression models in the background and ranks the top statistical drivers!

2. The Decomposition Tree: Ad-Hoc Root Cause Exploration

The Decomposition Tree visual allows executives to break down a high-level metric across any sequence of dimensions on demand:

The "AI Split" Feature:

Instead of manually choosing the next dimension, users can click the Sparkle Icon (+):

• High Value: Power BI automatically scans all dimensions and picks the attribute with the greatest positive contribution!
• Low Value: Automatically identifies the attribute driving the greatest decline.