Data clustering

When we make PowerBI reports with Scatter chart there is a technique where, by using machine learning mechanisms, data groups are grouped into clusters. Clusters are represented by different colors and in this way it is easier to see the connection between groups of data. Once created clusters can also be used to create other PowerBI visuals.

How is this technique applied? After we got the test data, suitable for creating a Scatter chart, will add this visual to the report. In our example, we analyse the sale of one of the five products to users depending on their age, so we should drag GODINE (AGE) dimension to x axis, PROIZVOD (PRODUCT) to y axis, and RBR into the Details field. For better detection, we can further set the range of values on the axes so that the data is displayed at the centre of the chart.

Once we have created the chart, open the menu at the top right corner and select the Automatically Find Clusters option. This will launch an action that, by using algorithms for machine learning, divides data into clusters by mutual similarity. Depending on the amount of data this operation may take longer or shorter. When finished, data will be coloured with different colours, depending on which cluster they belong to.

If you like, you can now use clusters to create other types of charts, for example, Column charts which brings one more insight into the data within the report.