Please see Part 1 of this tutorial for background on setting up the cumulative totals and rankings. This tutorial is a continuation that gets counts and proportions of top and bottom features. It then demonstrates how to build dynamic stacked bar charts and descriptive statistics of our top features and their influences on clicks.
Get Counts of “Top” and “Bottom” Features
Now that we know which features are “top features”, we need to know how many features are actually top features, and how many total features we have. This way a proportion can be established and we can use them in our bar charts.
Top Features:
Bottom Features:
Lastly we need the percentage of clicks generated, just as a number, of the top and bottom features. This measure is used in the KPI for the percent of clicks generated by the top features.
Top Features % measure:
Get the Bottom % of features measure:
For Percent of Features that are Top Features, use a 100% stacked column chart and drag the new variables (FeatureCount_Top, FeatureCount_Bottom) onto the Y axis. Power BI will compute them in a way that they turn into percentages. Format the colors, labels, axis, accordingly to your needs. I removed mostly everything except the one data label for Top Feature Percentage and also made them different colors so we pay attention to the purple section.
Copy/Paste that chart, or create a new 100% stacked bar, but this time, drag the FeatureUsage_TopFeature% and FeatureUsage_BottomFeature% onto the Y axis. Format the chart accordingly.
For the two KPIs above the charts, we just need to create a couple more measures. Percent of Features is computed as follows, and then drag it onto a KPI card. First get a Feature Count, which accounts for blank rows (e.g. a blank product category) that could potentially show up (as it can in my data), and also added a 0 so we can divide without errors and so that a “(Blank)” does not show up in the KPI visual:
Percent of Features — use this in the KPI Card:
Create the KPI Cards and place them above the stacked bar charts:
I used a final formatted measure to read out our statistics in a sentence format. This ties all of these measures together. Add it in a Card visual to get something like this:
The last visual is a Table that illustrates more statistics on the features. I chose to add the isTopFeature indicator next to the feature’s name for quick reference while the audience can also get some more specific information about each feature. Format the icons accordingly if you like.
We now have a pareto principle dashboard that is easily filterable so that users can evaluate changing categories, dates, etc. No calculated columns were created so the visualizations are extremely flexible and dynamic.
Be the first to comment