Machine Learning, BI and User Experience

D3-big_data_visualizations-500x338I’ve been friends with the CEO of an undisclosed BI company for some time and one of the things that our conversations have lead me to focus on is the idea of application heat mapping to look at how users behave and interact with an application. What is exciting is the innovative change, along with the orchestration engine and AI analytics of the IOE system, as an opportunity for how we do better application heat mapping and a great extension to the kinds of BI my friend’s company back east might be able to deliver to their users.

Heat Mapping

So what is an application heat mapping system? If you have heard of application analytics before you might have seen reports maybe exported to Excel or some pretty data grid that shows all these hit points in the application. The idea of a heat map in an application is looking at behavior; if the heat map is just looking at pages and hit counts on a given page or application view this is one common method of implementing a heat map.

Heat Mapping and UX

For an idea that allows better application of analytics, and presentation in a more consumable fashion, take this example interaction model; were we can see a specific flow through an application. From a visualization standpoint, this allows us to produce a report format that shows that spread sheet information in a more consumable and relevant method of looking at it.


Figure 1A - sample interaction model from AR game

Basically Humans can now ‘see’ the data that was in the report previously. So how do we take that to the next level you might ask?

Enter Smart Analysis

With tools like power BI and the information orchestration engine, with a touch of machine learning, we can start to extrapolate additional report data and visualize it in a method that actually is even more useful. Instead of just saying users went to this screen 55 times vs 1 for screen x we can start to draw conclusions about their behavior; for example, users went to screen x and paused for 35 seconds or more 95% of the time and 3 out of four times the next step is a help screen and we can draw a conclusion that this particular screen, while apparently popular, needs help.

See this example:



Figure 1B - sample interaction model from AR game with reporting

Using Business Intelligence, and then visualizing it to the right users, is a key component of the application of emerging technology, as well as better UX, to ‘better’ understand what is going on with our applications. In the second example, we actually use an application ‘map’ and overlay that with the data; so instead of a reference we see where the problem is visually as well as see other problems that we can then drill down into. This provides the visual context and, based on the current engine, is relatively straight forward to do. The best part is, now a number of companies I’m associated with want to do this kind of UX (User Experience) technology. We will see which one wants to land it first.

* hero image



One thought on “Machine Learning, BI and User Experience

  1. Business Intelligence is so relevant to companies today & I’m extremely happy that PowerBI took what was very complex in 2008 (I helped Lynn Langit with her book “Smart Business Intelligence with Microsoft SQL Server 2008”) and made it user-friendly. The other part very important to BI is using CLEAN data. Ask me how you can be guaranteed to have CLEAN data 🙂

Leave a Reply