The Analyze Feature and AI Visuals

Quick Answer: Power BI includes powerful AI features for automated analysis. The Analyze feature explains data changes automatically. Key Influencers finds what drives a metric up/down. Decomposition Tree explores root causes interactively. Q&A lets users type questions in natural language. These features are heavily tested on the PL-300.

The Analyze Feature

Explain the Increase / Decrease

Right-click any data point in a visual and select:

Analyze → Explain the increase  (or Explain the decrease)

Power BI will:

1. Analyze all related dimensions and measures
2. Identify the categories that most contributed to the change
3. Display a waterfall chart showing the decomposition
4. Rank contributing factors by impact

Example: If Q3 revenue increased significantly compared to Q2, the Analyze feature might show:

• Product category "Electronics" contributed +$500K
• Region "West" contributed +$300K
• Customer segment "Enterprise" contributed +$200K

Find Where a Distribution Is Different

For category-based visuals:

Analyze → Find where the distribution is different

Identifies which sub-categories have a significantly different distribution compared to the overall data.

Key Influencers Visual

The Key Influencers visual uses AI to determine which factors most impact a target value.

Setting Up Key Influencers

1. Add the Key Influencers visual from the Visualizations pane
2. Drag the target metric to the "Analyze" well
3. Drag explanatory factors to the "Explain by" well

Two Tabs

Interpreting Results

The visual shows statements like:

• "When Product Category is Electronics, Revenue is $45 more on average"
• "When Region is West, the likelihood of a High Rating increases by 2.5x"

Configuration Options

• Analyze by: Continuous value (analyze what increases/decreases it) or categorical value (analyze what influences it being a specific category)
• Expand by: Additional grouping for deeper analysis
• Count: Minimum number of data points for significance

Decomposition Tree

The Decomposition Tree visual provides interactive, AI-driven root cause analysis.

Setting Up Decomposition Tree

1. Add the Decomposition Tree visual
2. Drag the measure to the "Analyze" well
3. Drag dimension fields to the "Explain by" well

Interactive Exploration

At each level, you can:

• Click a specific category to drill into that branch manually
• Click the AI icon (lightbulb) to let AI find the highest or lowest contributing factor
• Expand multiple paths to compare branches

AI Split Options

Example: Analyzing revenue drop:

1. AI finds "Electronics" has highest revenue → click to expand
2. AI finds "West Region" has highest revenue within Electronics → click to expand
3. AI finds "Q4" has highest revenue → understanding seasonal pattern

Q&A Visual

The Q&A visual allows natural-language queries:

Insert → Q&A visual

Users type questions like:

• "What is total revenue by region?"
• "Show me top 10 products by profit"
• "Revenue trend for last 12 months"
• "Which customer had the highest sales in Q1?"

Power BI interprets the question and auto-generates an appropriate visual.

Q&A Setup for Authors

• Synonyms: Add alternative names for fields (Settings → Q&A Setup → Synonyms)
  • "Revenue" = "Sales", "Income", "Turnover"
  • "Customer" = "Client", "Account", "Buyer"
• Suggested questions: Pre-define questions for the Q&A visual
• Review questions: See what users are asking to improve the model

Grouping, Binning, and Clustering

Grouping

Combine discrete values into custom groups:

Right-click axis values → Group

Example: Group individual countries into regions: US + Canada → "North America"

Binning

Create equal-width ranges for numeric or date values:

Right-click numeric field → New Group → Bin type: Bin size

Example: Group ages into bins: 0-18, 19-35, 36-50, 51-65, 65+

Clustering

AI-driven grouping of scatter chart data points:

In a scatter chart → More options (...) → Automatically find clusters

Power BI uses k-means clustering to identify natural groupings in the data.

Reference Lines, Error Bars, and Forecasting

Reference Lines

Add horizontal or vertical lines to charts for context:

Error Bars

Show uncertainty or variability ranges on data points:

Format pane → Error Bars → Enable → Configure upper/lower bounds

Options:

• Fixed value, Percentage, Percentile, Standard deviation
• Custom fields (upper bound measure, lower bound measure)

Forecasting

Built-in forecasting for line charts:

Analytics pane → Forecast → Enable
→ Configure: Forecast length, Confidence interval, Seasonality

Settings:

• Forecast length: How far into the future to predict
• Confidence interval: Range showing prediction uncertainty (e.g., 95%)
• Seasonality: Auto-detect or specify period (12 for monthly, 4 for quarterly)
• Ignore last N points: Exclude recent data points from the model

Detecting Outliers and Anomalies

Anomaly Detection

For line charts with time-based axes:

Analytics pane → Find Anomalies → Enable

Power BI uses statistical models to:

• Identify data points that fall outside the expected range
• Mark anomalies with special markers
• Provide explanations for why each anomaly occurred
• Show contributing factors

Outlier Detection with Scatter Charts

• Use scatter charts to visually identify outliers
• Combine with reference lines (average, standard deviation) to highlight boundaries
• Apply clustering to separate outlier groups from normal data

On the Exam

The PL-300 frequently tests:

• Using the Analyze feature to explain increases/decreases
• Setting up Key Influencers and interpreting results
• Using Decomposition Tree for root cause analysis
• Configuring forecasting on line charts
• Understanding anomaly detection capabilities