Wednesday, February 4

Don't live in the past, predict the future

If Jethro Tull is "living in the past".  And hindsight is 20/20.  Then you're reading this blog in the moment to learn how to predict the future.  From a business intelligence perspective, that is.

Thanks to Accutate for providing this short article on Predictive Analytics.

"Predictive analysis can be an extremely useful tool for many different types of businesses. In fact, where there is any type of data warehousing there should be implementation of a business intelligence program that includes predictive analysis. However, in order to learn how your business can profit from this facet of business intelligence, you are going to have to understand exactly how and why predictive analysis works.  

The main idea of predictive analysis is to use current and past data to predict future events. The goal of the statistical techniques used in predictive analysis is to determine market patterns, identify risks, and predict potential opportunities for growth. In addition, data relationships can be reordered to determine the most plausible outcome of possible solutions and patterns can be recognized that might have the power to alter the outcome of a probable event.  

One of the most important aspects to reliable predictive analysis is data quality. The information provided by predictive analysis can only be as effective as the abundance and accuracy of data available. Data quality is absolutely necessary to the process of predictive analysis. In order to attain accurate business intelligence, companies must maintain quality data.  Predictive analysis requires both past and current data about many different things including customers, businesses, products, and the economy. All of this information is used to draw relationships and patterns between sets of data. If the data is accurate and well maintained, then the business intelligence produced will be high quality as well.   

In the past, predictive analysis was mainly used for newly emerging technologies. However, in recent years these practices have quickly started to become common for mainstream businesses. There are a few differences between the ways that these techniques are currently used and how they were used in the past. One of the main reasons for these differences is why companies use predictive analysis. In the past, these techniques were used for long-term analysis of market and consumer trends. However, in recent years, the mainstream implementation of predictive analysis techniques has tended to focus more on immediate, tactical uses. Because of the “real-time” nature of this business intelligence, more and more companies are using predictive analysis as standard in making predictions about particular industry markets and consumer trends.  

Some of the industries that have started utilizing these business intelligence techniques include telecom, insurance, pharmaceutical, and financial industries. All of the companies in these different business sectors have been able to use predictive analysis to make the right decisions to move their businesses in a positive direction. These processes can help with economic predictions as well as predicting the behavior of businesses and consumers. This type of information, made available in an efficient manner by business intelligence, is understandably invaluable. It can turn a simple prediction into intelligence that is more precise than even the most educated guesses. Predictive analysis with appropriate attention paid to data quality has made it easier than ever for businesses to make accurate market and consumer predictions and thus smarter decisions for the growth of their company."


Jim Franklin said...

This is a good description of what I call 'Quantitative Predictive Analytics'. When quality data is available, then this approach works. But what if the data does not exist? or is expensive or hard to get?

That's where Crystal Ball comes in. A good example is a new product launch. Because it is new, there is no data for product quality or the elasticity of demand. With Crystal Ball, you can optimize your decision, given the data you do have and the uncertainties that you face.

Qualitative Predictive Analytics is more the art rather than the science of PA. It puts more weight on the business logic that connect past data to future outcomes than on the data itself. I like to say, "Let the data guide you, not rule you". Sometimes the computer can make the decision, but more often than not a human will be making the decision.

Jim Franklin

Note: I am an Oracle employee and these opinions reflect my thinking and are not Oracle policy.

Tom Hudock said...

Jim, good point. Would I be right in assuming that since you're from Oracle, you are talking about Decisioneering's (now Oracle's) Crystal Ball product? From what I read, it's an Excel add-on with business rules under the covers for forecasting and "uncertainty management".

How would you compare this to other forecasting and planning products?