Human intelligence as main interface for BI

Never in the past were we able to collect such a vast amount of data. Technology these days is capable of collecting, storing and reproducing data in an affordable and performing way. The challenge in the near future will be to deal with all this information and what is more important, to find a way to use it to our advantage.  More and more businesses are involved in this data struggle as they have less structured and more complex data.  It is quite clear that the traditional way of looking at data in the form of reports and tables will not be able to master this new challenge. The only way to handle huge volumes of data is to use human capacities to understand pictures, forms and trends using our eyes and perception capabilities. Therefore data need to be transferred from texts and tables into visual forms.

DO I NEED TO DRAW YOU A PICTURE?

When people try to explain a difficult situation, they often quickly use pictures and drawings to bring insight and to direct an audience to a conclusion because images speak out and are more relevant than wording. That is also why engineering students receive sketch lessons at University, not to become art students, but to learn how to explain complex problems and of course the solution in a simple an understandable manner.

These two examples illustrate the power of visualization. Knowing the importance of visualization, it is strange that in our day-to-day work and job information is often presented in form of numbers and texts, and often distributed in tables and reports.

DATA SIGNALIZATION AND EXPLORATION

Visualization of data is used in two main areas of Business intelligence.

First of all, visualization is used for signalization. People will pay much more attention to messages that address the reptilian coping brain. This part of our brain doesn’t listen that good, and will only wake up when stimulated by food, sex or danger. As the first two aren’t really widely accepted in a work environment, we can focus on danger, alert. Visualization formats that address our danger reflexes are called signalization. It is used mostly in dashboards and scorecards. 

Secondly, visualization can be used as a guide for data exploration. Visual data exploration is a very effective way to dive into BIG DATA. Without visual data exploration techniques BIG DATA, for instance, quickly becomes useless.

SIGNALIZATION FOR BUSINESS INTELLIGENCE AT A GLANCE

What’s a dashboard? According to Stephen Few [1] , information and results presented at once and on a single screen are the core elements of a dashboard:

A dashboard is a visual display of the most important information needed to achieve one or more objectives which fits entirely on a single [computer] screen so it can be monitored at a glance.

- Stephen Few, Dashboard confusion - March 20, 2004

Based on this definition, we can understand how to best use visualization in dashboards for signalization alerts. Here’s a five key-rule for optimal dashboards:

1. Display data visually.  As explained above, humans interact in a visual way with information. Therefore, knowing how we interact is mandatory when designing a dashboard. People are very sensitive on colors (danger/safe/neutral) and their intensity, orientation, size, line thickness, enclosure and marks.  
2. Focus on important data only. Decisions are based on the most important data, meaning aggregated data. Too much details will only deflect people from the real data. Dashboards will only display relevant and highly aggregated data.
3. Achieve objectives. Dashboards will normally only display smart objectives. Here we need to divide the objectives in Result indicators and (Key) Performance indicators. Although result indicators can warn you to investigate further, they’re mostly too late and not focused enough. 
4. Fit one screen. People don’t scroll to find information. This pinpoints the importance of the limited canvas the dashboard can use. Every centimeter should matter. Reduce pictures, texts, information and all other fancy decoration in your dashboard and focus on data. Every pixel should handle data, or should be used to highlight the data. 
5. Just a glance. Dashboards are only looked at with a glance and with that glance we should react. It can be that everything is OK, no visual signs point to any attention. If this is the case because indeed everything is OK, then it is fine. 

VISUAL EXPLORATION FOR ANALYSIS AND DATA MINING

The purpose of data mining is not the in-depth analysis of massive data but to find correlation, links and trends between different data elements. Once we have discovered these links and trends, we can analyze the correlation and look at the data. But when presented in text form, the amount of data that can be overlooked by a person are only about 100 items. A drop in the ocean when dealing with data sets containing millions of data items. 

Thus a visual exploration process can be seen as a credible hypothesis generation process to understand large volumes of data. Visualizing data allows users to overlook, gain insight and to come up with new hypotheses on information contained and expressed by large amounts of data. The verification of the hypotheses can also be done via visual data exploration, but it may also be accomplished by automatic techniques from statistics. However, automated data mining is only possible if there is set of data that has some correlation and is cleaned from noisy data. Therefore the first step is to use visualization to filter and to zoom into relevant information.

Visualization techniques can go from simple X,Y plotting over bar charts and pies to more complex scenarios like geometrically-transformed displays, iconic displays, dense pixel displays and further on. Combining multiple techniques in one new one can also result in useful information. 

Follow this link to see an interesting exploration graph where color and positioning are used to give an immediate overview of relations between genes and detect neighbourhoods in mating response genes. 

SUMMARY

Data visualization is essential to Business Intelligence, for reporting and management as well as for analyzing and mining. The exploration of BIG DATA is a complex and difficult problem where a powerful tool is our own human eye capability of identifying forms, correlation and trends. And when visual exploration is combined with automated data mining techniques, it can reach its most important value.

If we can do this in a user-friendly way, and make it available on any device, not only will it be economically valuable and remove the pain points of complex data models, but it will also bring pleasure to the user.

The art in BI is to make complex things look simple. It starts with complex scattered data but should end with easy to oversee information. Our corporate BI simplexity policy is greatly about visualization.

Kris Bornauw , Sales Director / Expert BI, EoZen

[1] Stephen Few has over 20 years of experience as an innovator, consultant, and educator in the fields of Business Intelligence (data warehousing and decision support) and information design. He regularly teaches at conferences such as those presented by The Data Warehousing Institute (TDWI) and DCI, and also in the MBA program at the Haas School of Business at U. C. Berkeley. He is also the author of Show Me the Numbers: Designing Tables and Graphs to Enlighten, Analytics Press, 2004.