Loading...
You are here:  Home  >  Data Education  >  Big Data News, Articles, & Education  >  Big Data Blogs  >  Current Article

The Road to Prediction: Using Unstructured Data

By   /  October 14, 2016  /  1 Comment

Click here to learn more about author Bryan Bell.

Prediction is tough and rarely a guarantee. If we were able to predict the future, there would be no Vegas, or at minimum a much poorer version. Weather would cause fewer surprises and the World Series, the Davis Cup, the Stanley Cup, the World Cup, the Super Bowl, despite the not so predictable commercials, and the Daytona 500 would not be nearly as interesting to watch. In the world we live in, there are so many variables, nuisances and unknowns, it is difficult at best, to predict a specific outcome or what the future may hold. If we were able to consistently and accurately predict the future, it would be similar to having tomorrow’s Wall Street Journal, and those with access, would retire rich the day after tomorrow.

What we have become very good at, is collecting information from the past, identifying trends and better understanding what may happen in the future. It is well known that the best predictions come from analyzing the past, accurately interpreting what has occurred, recognizing repeating events and identifying trends to better understanding cause and effect and only then predicting an anticipated outcome. To continue moving toward accurate predictions, we must continue to increase our ability to not only collect information, but accurately interpret the vast volumes of unstructured data in a contextually correct way and establish links, cause and effect. Lacking context, is like hunting and knowing little about what the animal looks like; other than it is usually within 20 meters of water and is often found under a tree. I ask, how precise can one be with a prediction, if we cannot be precise in how we interpret the information we are using to make the prediction. If we cannot accurately interpret information, how can we expect to understand what may or may not happen in the future?

Back to the weather, it is a common experience to check the weather prior to heading out around town, preparing for an event or departing on a long-distance trip. When I was a teenager, predicting the weather seemed to be more about luck and superstition than science, based on personal observations, the Farmer’s Almanac or the Magic 8 Ball. But even if tools for weather prediction were not perfect, there were still many tools in place to enable monitoring and measuring and therefore better understanding the future by a better understanding of the past.

As time has passed, new data monitoring and measuring tools now allow meteorologists to create complex models that enable very predictable weather forecasting. And while predicting the weather is still not perfect, we are much closer than seemed possible even 10 years ago. I can’t help but see the similarities between this and the business world. Companies have been collecting data for years and most are quite good at explaining what happened in the past. Today, with the advance of tools and technologies that are able to capture, store and analyze what was once uncollectable, companies have now become progressively better at using it for decision making and forecasting. Here, I am obviously talking about the vast volumes of unstructured information. But, even with these new rich sets of available information, there is still a lot of information left on the table, unused. Today’s information can be collected partly due to the many data repositories now accessible by organizations and / or being actively used within organizations, (i.e. the Internet, CRM systems, social media, market research verbatim, etc.). And thanks to artificial intelligence tools, cognitive computing such as semantic technologies it can be understood and structured in a format that can now be used to support new innovative predictive models.

Using the example of Internet content, all of the things embedded in text—topics, tone, style, relationships between concepts, etc.—when used as attributes/variables that can be associated to each piece of text to describe context and meaning to better understand content. A predictive model can now take all of this information into consideration across many dimensions and can have a huge bottom line impact. For example, understanding what is causing churn in your customer base, whether it is time to launch a new product and what the product should be, where to drill the next oil well, invest in research to target a specific disease or who is likely to be the next fatality within a city neighborhood. These are all complex matters to predict and for those who investment in understanding and interpreting information in the best ways, will make the most progress in the shortest amount of time, realize huge benefits and capture the advantage.

The good thing is, you don’t need a major capital investment to put such systems in place. And, once deployed, these solutions will help pave the road for both short-term and long-term success.

Today, more than any other time before, businesses really do have a level playing field. And everyone has the opportunity to be the next member of the Dow or the next weather geek, perfecting the art of always knowing what to pack for the next trip.

About the author

Bryan Bell Bryan is Expert System Enterprise EVP, Market Development. An industry veteran, Bryan has extensive experience in search, automated metadata extraction, taxonomy creation, knowledge management and semantic technologies. Prior to his tenure at Expert System, Bryan contributed to the growth of Teach.com, Autonomy, Smartlogic and Concept Searching. Bryan holds a B.A. in Communications from Murray State University.

  • “What we have become very good at, is collecting information from the past, identifying trends and better understanding what may happen in the future. It is well known that the best predictions come from analyzing the past, accurately interpreting what has occurred, recognizing repeating events and identifying trends to better understanding cause and effect and only then predicting an anticipated outcome.”

    It is true that looking at the past is an important part of making predictions for the future. Despite our best efforts, the future can’t always be predicted, but identifying trends is still valuable because it gives us a clear idea of what is possible.

You might also like...

Machine Learning Algorithms: Introduction to Random Forests

Read More →