Sophisticated use of data analytics marks the moment when the ”Information Society” finally fulfills the promise implied by its name. According to Wikipedia, there is however currently no universally accepted concept of what exactly can be defined as an information society and what shall not be included in the term. Most theoreticians agree however, that a transformation, that formed most of today’s net principles and currently as is changing the way societies work fundamentally, can be seen as started somewhere between the 1970s, the early 1990s transformations of the Eastern Europe and the 2000s period.
Professor Frank Webster, PhD, author of the book, ”Theories of the Information Society”, has listed five major types of information that can be used to define information society. These are:
– Technological Information
– Economic Information
– Occupational Information
– Spatial Information
– Cultural Information
Professor Webster states, that the character of information has transformed the way that we live today and how we conduct ourselves centers around theoretical knowledge and information. The data takes centre stage. All those digital bits, that we have gathered, can now be harnessed in novel ways to serve new purposes and unlock state-of-the-art forms of value. But this requires an open-minded way of thinking and will challenge our companies, institutions and us. The only predictable thing that is certain is the fact that the amount of data to be handled and the needed power and data storage to process it all will continue to grow in sync.
Most people have considered data analytics principally as a technological matter focused on the hard- and the software used for processing the data. With increasing quantities of data, it has become increasingly important to decide, what insights you want to gain when trying to separate the wanted signals from the surrounding noise. We also believe that more emphasis needs to be shifted to issues, like what needs to be done when the data has spoken.
Instead of being obsessed about the accuracy, exactitude, cleanliness, and rigorousness of the data, we can today face the data from a more liberal angle and give some slack. We shouldn’t however accept any data that is outright wrong or false, but some messiness may become acceptable in return for capturing a far more comprehensive set of data and getting new intelligence about issues residing below the surface. In fact, in some cases big and messy data sets can even be beneficial, since when we tried to extract knowledge by using just a small, exact portion of the data, we ended up failing to capture the breadth of detail where so much knowledge lies.
The main idea in many cases is, that it may be more advantageous to focus on clarifying what information is hiding inside of the data instead of trying to solving the why -issue. There might be multiple causes for certain phenomena when solving skills related problems. As an elegant way for finding insights form the data we approach it considering in our analysis its group-related statistic characters. They describe the stochastic relationships within groups, but never allow explicit conclusions to be made about individuals of the population or about the causality of single incidents. This doesn’t however reduce the empirical relevance of the insights.
Because correlations can be found far faster and cheaper than causation, they are often preferable. But for many everyday needs, knowing what is not good enough. Stochastic data correlations can however point the way toward promising areas in which to explore causal relationships. You need to know, where to dig deeper for answers for the question: Why?
Of course, causality is nice when you can get it. The problem with real life multi factor causality is, that it is often hard to expose causality, and when we think we have found it we are often deluding ourselves. Wishful thinking can possibly lead you to wrong avenues of research. A fundamental reason for that is, that when we have more data available and since more aspects of the world are being datafied and collected, there is a danger for us to go astray.
Much of the value of data will come from its secondary uses, its option value, not simply its primary use, as we we’re accustomed to think about it. As a result, for most types of data, it seems sensible to collect as much as one can and hold it as long as it adds value, and let others analyse it if and when they are better suited to extract its value.
Sometimes important assets will not just be plainly visible pieces of information. The bulk of data created by people’s interactions with skills and learning is a source which a clever company can use to improve existing analytic services and even launch entirely new ones. Because we are now able to predict how different course scores may influence the future of our sales or the efficiency of our services, this will allow us to take remedial steps to prevent problems or to improve outcomes. We will detect students and performances which will start to slip before the expiry dates of the courses. We can support those, who have problems, and those who struggle. Successful supportive action will improve the performance of the individuals and be beneficial for the company.
Nothing is preordained, because we can always respond and react to the information we receive. Calculated predictions are not set in stone – they are only likely outcomes, and that means that if we want to change them we can do so. When we are able to see problems arise in the horizon, we are able to rapidly take corrective actions. Because we can never have perfect information, our predictions are inherently fallible. This doesn’t mean they are wrong, only that they are always incomplete. It doesn’t negate the insights that big data otters, but puts learning analytics in place – as a tool that doesn’t offer ultimate answers, just good enough ones to help us for now with our everyday business challenges until better methods and hence better answers come along.
Written by Kari Hartikainen, Sales, Boudin Oy