This is additional content for the book, “Data Analysis with Microsoft Power BI.” It fits in Chapter 6 – Basic Data Visualization in the middle of page 168 prior to the Other Visual Elements section.

Key Influencers Visualization

The Key Influencers Visualization is used to determine which attributes are most strongly related to a particular characteristic or a particular change in a measure. This visualization uses an artificial intelligence (AI) algorithm behind the scenes to make this determination.

Using a Key Influencers Visualization with a Characteristic

1. Click the yellow tab containing the plus sign to add a new report page.
2. Right-click the Page 1 tab and select Rename from the context menu.
3. Replace “Page 1” with Home Owner Key Influencers and press ENTER.
4. In the Fields pane, expand the Customer table.
5. Check the “Home Owner?” field.
6. In the Visualizations pane, click the Key Influencers visualization icon as shown:

   

   The beginning of a key influencers visualization is created on the report page.

7. In the key influencers visualization, change the dropdown from “N” to “Y” as shown:

   

   We are asking the key influencers visualization to tell us the best predicter of the “Home Owner?” field having a value of “Y”.

8. Check the following fields in the Customer table:
   – Married?
   – Number of Cars Owned
   – Number of Children at Home

   These fields will be added to the “Explain by” area in the Visualizations pane as shown:

   

9. Use the sizing handles to enlarge the key influencers visualization to it fills the entire report page.

Analyzing the Key Influencers Visualization

The key influencers visualization uses an analytics algorithm to look at each of the fields we placed in the “Explain by” area. It determines which field values have the strongest correlations with the desired value, “Home Owner?” equals “Y”. Those field values with the strongest correlations are displayed in the visualization.

We can see owning two cars is the biggest influencer (best predicter) of a customer being a home owner. This is followed in significance by having one to two children at home and by being married.

1. In the key influencers visualization, click the “Number of Cars Owned is 2” entry. (You can also click on the circle across from this entry.) A column chart appears on the right side of the visualization. This column chart shows how the data breaks down by “Number of Cars Owned” as shown here:

   

   The columns on the chart show the percentage of records in that category with the desired characteristic. In this case, owning a house. The colored column shows the data for the item selected on the left side of the visualization. Here, the colored column tells us 65.12% of the records where the “Number of Cars Owned” field is 2 also have the “Home Owner?” field set to “Y”.

   The black columns provide the data for the other item in this category. In this case, owning zero cars or owning one car. We can see that a much smaller percentage of records where the “Number of Cars Owned” field is 0 or 1 also have the “Home Owner?” field set to “Y”. The dashed line shows the average of the black columns.

   The number inside the circle shows the increase in the occurrence of the desired value that is obtained from the selected value. Here, owning 2 cars leads to a 1.44 times increase in the “Home Owner?” field being “Y” (rounded to two decimal places).

2. Click the arrow in the upper left corner of the column chart to hide the column chart.
3. Try clicking the other key influencer entries as well to view their column charts.
4. Click the Top Segments tab at the top of the visualization. A scatter chart appears as shown here:

   

   The AI within the key influencers visualization used a segmentation algorithm to divide the records into a number of groupings or segments. Each segment has a certain set of characteristics. Here, five segments are shown. More that five segments are created, but only those segments with the greatest correlation to the specified criteria – having “Home Owner?” equal “Y” – are shown.

   The higher the segment’s circle on the chart, the larger the percentage of records in that segment meeting the specified criteria . The larger the circle, the more records there are that fall into that segment. We can see that 100% of the records in Segment 1 have the “Home Owner?” field equal “Y”. We can also note that Segment 1 has a smaller population relative to the other segments.

5. Click “Segment 1”. (You can either click on the “Segment 1” label along the bottom of the chart or on the corresponding circle.) The segment criteria appears as shown here:

   

   The left side shows the criteria that define records in this segment. In this case, all records in this segment have “Married?” equal “N”, “Number of Cars Owned” equals 2, and “Number of Children at Home” is greater than 2. The right side shows the percentage of records in this segment matching the desired criteria compared to the overall population. It also shows how the size of this segment compares to the overall population.

6. Click the other circles at the top of the visualization to see the characteristics of the other significant segments.
7. Click the “X” in the upper right corner of the segment characteristics to return to the segments scatter chart.
8. Click Save.

Using a Key Influencers Visualization with a Measure

1. Click the yellow tab containing the plus sign to add a new report page.
2. Right-click the Page 1 tab and select Rename from the context menu.
3. Replace “Page 1” with Sales Key Influencers and press ENTER.
4. In the Fields pane, expand the Orders table.
5. Check the “Sales in Dollars” measure.
6. In the Visualizations pane, click the Key Influencers visualization icon. The beginning of a key influencers visualization is created on the report page. Note the dropdown in the key influencers visualization is set to “increase” as shown:

   

   This indicates we want to know what are the key influencers that cause Sales in Dollars to increase.

7. In the Fields pane, expand the Customer table and check the State field. A message appears in the key influencers visualization as shown here (you may have to expand the visualization to see the message):

   

   A message will appear whenever the analytics algorithm cannot be processed successfully.

8. In the Fields pane, expand the Promotion table and check the Promotion field.
9. In the Fields pane, expand the Sales Person table and check the Sales Person field.
10. In the Fields pane, expand the Store Type table and check the Store Type field.

    Because we are working with a measure this time, the analysis is being done at an aggregate level. Because of this aggregation, the AI is not finding any key influencers. We need to perform the analysis at a more granular level to get meaningful results.

11. In the Fields pane, expand the Product table.
12. Click the Product field to drag and drop it in the “Expand by” area as shown:

    

    The analysis is now done at the Product level.

13. Expand the key influencers visualization so it covers the left three-quarters of the report page. The report page should appear as shown here:

    

    The largest influencer found was having no promotion. The majority of our sales are not associated with a promotion, so this result is not very helpful. Let’s filter out the blank promotions.

14. Expand the Filters tab.
15. In the “Filters on this visual” section, expand Promotion.
16. Check “Select all”.
17. Uncheck “(Blank)”. We now have some more interesting results. The Filters pane should appear as shown here:

    

18. Collapse the Filters tab.

    NOTE: By using this filter, we are removing all sales that were not done on a promotion from the analysis. What we are really determining now is, what are the key influencers that increase sales where there is a promotion involved. To do a complete analysis, we would probably want to create another key influencers visualization where Promotion is not included in the “Explain by” area.

19. Click the blank area of the report page to deselect the key influencers visualization.
20. In the Fields pane, check the Product Type field in the Product table.
21. In the Visualizations pane, select the Slicer visualization.
22. Size the Product Type slicer appropriately. Do not select any item in the Product Type slicer. The report page should appear as shown:

    

23. Use the Product Type slicer to explore the key influencers on sales of each product type.
24. Click Save.