True Analytics as Seen by John Tukey & Benjamin Bloom


True Analytics as Seen by John Tukey & Benjamin Bloom

So who is John Tukey and who is Benjamin Bloom? They are two guys who know what “analytics” truly is, and, is not. Tukey is famous for the signal processing technique called the Fast Fourier Transform and the (original) Box Plot used in Statistics. The box plot idea was described about in his 1977 book called Exploratory Data Analysis. In addition he commented that there was too much post statistical data analysis being done and not enough exploration of the data (EDA) in raw form. Benjamin Bloom headed an educational research team which in 1956 produced a report known as Bloom’s Taxonomy. It listed six hierarchical levels of the cognitive domain, one of which is called Analysis. He defines analysis as the process of breaking the information into parts in order to explore and make inferences. Together, Tukey and Bloom are describing “true analytics”.


So what is “Predictive Analytics”? Well, in Bloom’s taxonomy, the level above analysis is called “synthesis” and the level below is called “application”. With synthesis one builds a structure or pattern from the elementary parts and forms a plan. With application one solves problems based on acquired knowledge. For now, we will say that Predictive Analysis is a combination of the two. The point is Predictive Analysis is not Analysis, at least for the sake of argument in this blog.


So where are we going next? Well, now that the focus is on the elements information, we need to see them and for lack of a better term, (ready?), to visualize them. I came across a catchy title to a paper being presented at the Data Visualization Innovation Summit held this April in San Jose. The author is Elijah Meeks and his paper is titled ‘Beyond Line and Pie Charts: Practical Applications of Complex Data Viz’. Although I am not sure exactly how focused the paper is on visualizing analytics in a form more pure than prediction, the title does suggest taking a deeper look. For me, visualization of pure analysis is as deep as the number of pixels you have on your computer monitor. Again, like Elijah Meeks, I am thinking beyond pie charts, histograms, plots and tables. Give me an X-Y axis where all I need is one pixel to indicate a fact.


Let me explain. Let’s say your screen has a resolution of 1000×1000 pixels. Now let’s also say you have a color resolution of 256. That means there are 256,000,000 ways to ‘color the facts’ with pixels. So, if you had a million customers, and you could enumerate the facts to some 256 level scale, you could see the “lay of the land”, so to speak, of all 1,000,000 customers at once with a range of 256,000,000 facts, easily. For the remainder of this paper, let’s call this level of visualization by a new name, one that I recently made up but one connected with communications. We’ll call it Base-Band Visualization but I am open to variations. The point here is that we are looking at information in a basic form; a form in which no higher level synthesis or application has been applied, yet.


Where is this to be used? In your Business-Information Technology Eco-System (BIT Eco-Sys). You see, the development of quantitative diagnostic modeling tools can in fact be inspired by an iterative process that begins with BIT Eco-Sys dialogue. The argument here is that with “true analytics”, evolutionary algorithms on real-time data will benefit greatly from a pre-emptive strike presentation of base-band visualization to the BIT Eco-Sys team. The combination of domain professionals, IT specialists and data scientists together can then construct relevant solutions that shape a client’s business. Then the team can demonstrate how the engineered algorithms depict solution intensity and deviations perceived by the client during the process, collecting the data in a useable form magically emerges as a solution that is easy to assimilate into a client’s business structure. And, incidentally, not only business structure, but data warehouse structure, i.e., true analytics is data warehouse agnostic. Whew!


Bottom line, the better your analysis via Tukey and Bloom, the better your predictive, discriminate and interactive relationships development for predictive analysis. Curiously, I have not included one visual; I have instead relied on your creativity that resides in you. If you would like to see examples please check out my previous blogs: BLOG1BLOG2, then contact us to talk.


Have a question for our experts? Leave us a comment below, check out our Big Data page or contact our team directly at at

No Comments

Post a Comment


five × three =

Stay Updated!

Sign up to receive our latest blogs and news. Join our community to receive our content first.