How can AI-driven features in Power BI improve the quality of insights

Power BI is a powerful business analytics tool that helps companies make smarter decisions. Recently, its AI features have become a hot topic. This blog will explore some of the key AI-driven features in Power BI, focusing on the Q&A functionality, predictive analysis, and AI visuals.

AI-Generated Visuals

Power BI includes two important AI-powered visuals that allow users to gain insights with minimal effort.

Key Influencers Visual

The Key Influencers visual helps users understand how different factors impact a specific metric in their data. Simply, it analyzes the data to identify the most influential factors and explains how these factors contribute to increases or decreases in the chosen metric. For example, the Key Influencers visual can show you if you want to know what affects a healthy nutrition score.

To create a Key Influencers visual, select the Key Influencers icon in the Visualizations pane. Then, move the metric you want to investigate into the “Analyze” field. For instance, when analyzing a nutrition score, the tool might reveal that a diet rich in vegetables is a significant positive influencer, while a low hemoglobin (Hgb) diet is a negative one.

Let’s take a look at some examples:

Vegetables as a Positive Influencer: The visual might show that including vegetables in a diet increases the likelihood of having a “Healthy” nutrition score by 3.92 times. A bar chart could also reveal that the percentage of “Healthy” scores is much higher when vegetables are part of the diet.

Low Hgb Diet as a Negative Influencer: On the other hand, a low Hgb diet could be identified as the top negative influencer, increasing the likelihood of an “Unhealthy” nutrition score by 6.86 times. The absence of meals and snacks might also contribute to an unhealthy score.

The visual can also break down the data into segments:

Top Healthy Segments: It might identify three key segments where nutrition scores are rated as “Healthy.” For example, two segments might have a 100% healthy score, while another segment might have a lower percentage but still maintain a majority of healthy scores.

Top Unhealthy Segments: Conversely, the tool might highlight segments with a high percentage of unhealthy scores. For instance, one segment with a small population might have the highest percentage of unhealthy scores, while another larger segment might also have a high percentage, suggesting the need for targeted nutritional interventions.

Decomposition Tree Visual

The Decomposition Tree visual allows users to drill down into multiple data dimensions to explore the root causes of specific outcomes. This visual requires two main inputs: the metric to analyze and the dimensions to explore.

Once you drag the metric into the field, the visual updates to show the aggregated measure. For example, you could analyze the count of patients with specific medical conditions and explore the root causes. The Decomposition Tree can work in two ways:

• High Value: It looks for the dimension that provides the highest value of the measure being analyzed.
• Low Value: It examines the dimensions to find the lowest value.

Q&A Feature

Power BI’s Q&A feature allows users to explore data by asking questions in natural language, such as “What was the total systolic BP change over time?” The tool then provides the answer in the form of a suitable visual, like a line chart.

You can add the Q&A feature to your reports in several ways:

• From the Visualization Pane: Select the Q&A icon.
• Double-click: Click into the background of your report to open the Q&A box.
• Insert Tab: Use the Insert tab to find the Q&A option, along with other AI visuals.
• Buttons Tab: You can also create a smaller version of the Q&A feature using the Buttons tab.

Once added, the Q&A feature can provide suggestions for questions, and users can convert the Q&A visual into a regular visual for further editing. There’s also a gear icon in the edit view that allows users to adjust the natural language engine for more precise answers. However, a limitation is that Power BI currently only supports English for natural language queries.

Predictive Analysis with Power BI

Power BI’s predictive analysis feature leverages machine learning to forecast future trends based on historical data. This capability is particularly useful for businesses looking to make proactive decisions.

Forecasting

Power BI includes built-in forecasting tools that predict future trends, such as revenues, customer growth, or inventory needs, based on past data. Users can adjust the forecasting model by changing parameters like forecast length, confidence intervals, and seasonality. This helps businesses make informed decisions, manage resources, and minimize risks.

AI Insights

Power BI’s AI Insights feature integrates with Azure Machine Learning, allowing users to apply machine learning models directly to their data. Users can either connect to pre-trained models or create new ones in Azure.

Anomaly Detection

Anomaly detection in Power BI identifies unexpected changes or deviations from the expected data patterns. For example, if you’re analyzing a gestational diabetes dataset, you could use the “Find Anomalies” feature to detect any outliers. By adjusting the sensitivity, users can control how easily the tool detects anomalies.

A limitation of anomaly detection is that it only works with line charts that have time series data on one axis and at least three data points on the other axis. The line chart also does not support secondary values, multiple values, or legends.