Maps are one of the most visual elements of Power BI that people love to see in their reports. Maps can show so much information in a short time because the Context is often very clear to the audience. In our Food Hygiene report for England, we have an ideal example of the limitation that Power BI has. This limitation is often used to refer to Data Capacity, rather than the truth that it is a visualisation layer limit. Visualisations can not typically show more than approximately 3,500 data points. For most visuals that’s not an issue; where it becomes important is in Maps and Scatter Plots. Microsoft has addressed this limitation by improving their sampling algorithm, so high-density areas appear densely populated, but lower populated areas also get “prioritised” in the visual, so they do not disappear.
If you are going to present scatter plots or Maps with more than 3,500 data points, you should therefore think hard about how that will work. Maps in Power BI can also only act as Scatter Plots in terms of the Algorithm processing (so low-density areas are not lost) if you use Lattitude and Longitude. Using other address information will trigger a look-up so the first unique 3,500 values will be used. In the video, we demonstrated this as the Lattitude and Longitude map loads (quicker) and showed areas of low density (data quality issues) while the postcode version only shows a small area around London.
The fix here is to create a Geography dimension that will standardise and group regions together, that may be a shapefile (and corresponding field list) or a postcode to Lat Long lookup. Using these can enable you to fix the data in Power Query or your Dataflows – Remember, the closer to the data you are when you make modelling changes, the lower the impact will be on your consumers.