What kind of Analytics ?

In various discussion with my customers and colleagues. I experienced a very controversy discussion around analytic.

Some understand the „what has happens,” also called root cause, others want to predict the future, meaning „what will happen,” a third group tries to answer the question, „what could happen“. Funny enough often is the answer a combination out of a combination of these. When it comes to our very fast decision making and consuming society this can cause some friction.
In the first case, lets call it descriptive analytic, it is the reason many companies invest money to avoid that, but after the incident, not predicted, happens.2434048_300dpi



Often you find it difficult to get to the root cause, since simple the data or information is no longer available to get to the bottom. In the 2nd case, statistical model show the business based on historical information what can get wrong, lets call it predictive analysis. In the last case, data scientists use machine learning to find the possible future, based on different information streams, it is known as prescriptive analytic. Like in chess where in the first move all possible is and there some prediction could be made, based on the information of the players, over time when more moves have taken the information and moves getting clearer.


The human brain is capable of doing this kind of prescription on the fly, we call that experience. By the way a reason why often young managers or entrepreneurs fail, but this is worth another blog. The key is here that the brain has build, by and expert, enough information and channels that the brain could do both predictive and prescriptive.
Having the scientists build the model, means the mathematical representation of the information combination, it can lead to a variety of possible outcomes, we call the data science or machine learning. Key is the the amount of information, it deep and historical very long. This is also on of the boundaries.


Many CIO and CTO from companies I talk today do not keep the data, for many reasons, mostly cost. So what can a Data Scientist than do ? Simple he build the model and than run it. Over the time it will be better, preserving the information. Like in the chess where you can see the extreme professional players analyze and predict movements in depth untrained could not imagine , or even provoke movements. This is not yet in this science embedded.


A couple of week ago I was fortunate to see a start-up and talk to the founders. It was all about customer intimacy and combining information in a company to serve a better customers relationship. I would drive this one dimension deeper. Why not using this as an internal knowledge base. Let´s predict what the corporate user needs to find around a specific subject, not only internal sources also external can be used here.


I am still wondering why HR representatives and headhunter not use predictive analysis to identify the right candidate for a job?
Yes you’re right the information in a company is often so filed that we will not have the possibility to combine them. Here we go with the concept of a data lake, building one repository of information and use it for internal and external benefit.

prop-jet engine

However the last topic is the „what could happen“ case, the descriptive Analytics. This is all about possibilities and often risks. I guess this is used in Six Sigma, and also man kind is used to do that. The mother warns the child about the possible accident, when walking on the street. The company strategic plans is all about that scenarios. I think this method means „I do not have enough data to be more precise. „


So what to do? Keep all the data, maybe randomized and anonymized, but keep it, because you never know what business can be build out of the treasure the company has generated over years.


Welcome in the Age of Information !