After having discussed creating clusters in Power BI using scatter chart, we introduce you to clustering using tables in this blog post. Clustering is one the simplest yet powerful AI feature in Power BI. Clustering helps you discover the natural grouping in your data set before you dig deeper into details. Power BI allows you to cluster your datasets by finding meaningful similarities in your data. Power BI intelligently detects clusters in data and also allows you to specify clustering requirements.
Let’s see what clustering looks like:
In the raw state, it seems nearly impossible to separate out objects based on their physical appearance. This is when clustering comes into play. Clustering groups the objects based on similarity between the objects and dissimilarity with others.
In the real life sales data set, there’s are a lot of products in our data and by just looking at the data, it is not easy to categorize the products based on metrics such as revenue and profit margin.
In this blog post, we will discuss how to discover this hidden clusters in your data and find the meaningful relationships in the data.
Clustering can also be performed in Power BI using a Table Visual.
Clustering can be performed in Power BI with:
- Scatter chart visual.
- Table visual.
- Custom visuals.
- DAX
In this blog post, clustering using scatter chart has been discussed.
For this blog post we will be using a simple sales dataset. The data model consists of a fact table (Sales_Table) and three dimension tables (Product_Table, Country_Table, Date). An additional virtual table contains the measures that will be used for clustering.
Before we move any further, lets define the business question under consideration
- What are the different categories of products in my raw data?
To perform clustering using the Table Visual:
Let’s first create a table visual
1.Click on the Table Visual in the Visualizations Pane.
Drag and drop the required fields to the field pane
2.Drag and drop the Product from Product_Table to the Values
3.Drag and drop the Revenue from All Measures to the Values field.
4.Drag and drop the Profit Margin from All Measures to the Values
Lets generate clusters
5.Click on the three dots at the top corner of the table visual.
6.Click on Automatically find clusters.
A pop-up menu appears. We will use the default values for this exercise.
7.Click OK.
Finally, analyze the clusters
8.Drag and drop the newly created field Product (Clusters) 2 to the Values field of the Table Visual.
9.Add the newly created field Product (Clusters) 2 in a slicer.
10.Click on Cluster 1.
Six clusters have been automatically detected by Power BI using Enhanced Artificial Intelligence.
Application of Clustering:
Some other practical applications of clustering include:
- Heath care analysis: identify the group of patients with similar symptoms.
- Biology: identifying groups of animals and plants with similar features.
- Marketing: identifying customers with similar behavior.
- City planning: identifying houses according to their value, type, etc.
- Earth quake analysis: identifying the dangerous zones.
- Insurance analysis: fraud detection and identifying clients with high claim cost.
Conclusion:
Clustering is an algorithm to group the data based on hidden relationships. These relationships are hard to understand by just looking at the data. Power BI however, intelligently detects meaningful relationships and creates clusters in the data. Clustering can be performed using scatter charts (discussed in the previous blog), using tables, custom visuals and DAX. In this blog post, clustering using the table visual has been discussed.
Are you a data analyst and want to learn more about Power BI?
Why not sign up for Power BI training in Australia. We provide our services in the following regions:
