GENERAL

Complexity versus Simplicity; Confusion versus Contribution, Impressive versus Knowledge and Understanding

The BIS.Net Team BIS.Net Team

Today, perhaps as always and perhaps forever there seems to be a tendency towards the first in each of the above pairs of nouns (and adjective).

One must watch TV medical shows to recognize how terminology is used to demonstrate medical knowledge and understanding. The objective is for the TV audience to be impressed by the medical knowledge of the ‘hero’ Doctor who invariably cures the patient. However, the TV doctor is not a real doctor, just an actor repeating scripts by the episode’s writers. The actor knows little nor understands real medicine. Medical TV shows thus demonstrate that a person who speaks fluently, using highly technical terms does not necessarily understand the topic he is talking about.

The practice of impressing with spoken and written words to hide lack of knowledge is evident in all walks in life. Many of us know that there were some students during high school years that were impressive in class, using big words, but performed poorly in exams, whereas the quiet unassuming students, who had nothing to prove performed at the highest levels. The assumption by those that make use of impressive jargon is that it demonstrates knowhow. The reality is that it demonstrates knowing all the right words, but it does not demonstrate understanding.

This is very relevant in the area of data analysis where there is an abundance of impressive jargon which raises expectations to unreasonable levels resulting in disappointment and loss of credibility of data analysis and science

Academia has also fallen into the trap of overcomplicating. Of course, academic thinking is at a higher than normal level of thinking, at a level of complexity that the average person will not understand. Academia must have certain standards of writing academic papers, with due reference given to previous work. Yet, it is difficult to deny that many papers have been written to convince the reader of the writer’s intellectual prowess, without really demonstrating subject knowledge. These types of papers tend to focus on squiggly symbols and complex jargon instead of ease of understanding. If more focus were placed on ease of understanding, then more academic research would reach practical applications in the real world in not just the analytics area. This situation can only change once it is understood that smart people make something complicated simple, not the other way around.

The situation of impressing is similar in the area of analytics and related areas outside academia in the ‘real-world’. In this instance impressing is achieved through hype designed not to build up the person, but the technology. The above image hypes AI by comparing AI with the human brain which is far more complex. Today we hear a lot about Artificial Intelligence, Machine Learning, Predictive Analytics. Machine Learning and AI can be based partially or completely on Neural Networks which is also hype designed to impress leading the average person believe that it is as powerful as the human brain. Yet, there is nothing magical about Artificial Intelligence, Machine Learning and Predictive Analytics. All three simply based on algorithms developed by human beings that simulate intelligence and learning. They are as good as the creators of the algorithms, which amount to no more than computer instructions. The computer does not learn, nor is the computer intelligent. The result of the hype means that there are many attempts to apply the technology by throwing say Machine Learning at a problem with questioning or asking for evidence that the technology will bring the promised results. The resulting failures, and there are many, discredit data analysis and data science.

Predictive Analytics is currently highly hyped. One would assume that this is about highly developed technology with which we can predict a future outcome, such as sales, crime rates, global warming rates, university student retention rates and more. The hypothetical average person, through the hype believes that new technology is applied with powers previously not possible. This is not always the situation. Sometimes the predictive analytics tools consist of tools used for decades, such as cluster analysis and regression analysis. More often it is no more than visualization, which in turn is a hyped version of basic charts such as a pie chart. Often the outcome is just the status quo, without real predictions.

Irrespective of the hype data analysis has remained in a time capsule. Hyping causes more harm than good as it eventually discredits data analysis and science. Hyping is about impressing and raising perceptions beyond reality. This is sad, because there are so many opportunities to bring data analysis into the future using the power of a computer and clever algorithms. To achieve these opportunities requires and intelligent approach where we do not resort to hype to build up technology but truthful communications of limitations so that these limitations can be addressed and so that the correct technology is applied to the problem.

Read the next article in the analysis of variation (ANOVA) series?