• Votes for this article 3 people voted for this
  • Dashboard Insight Newsletter Sign Up

Predictive Analytics
Reading the Tea Leaves of Business

by Giorgio Grosso, Business Author, Dashboard InsightTuesday, November 20, 2007

Is it possible to predict the future performance of an organization based on its past performance? Well business analysts who use Predictive Analytics certainly think so, and with good reason as their predictions are frequently correct. Predictive Analytics encompasses a host of techniques ranging from statistics to data mining, and uses a process that combines current and historical data to make informed “predictions” about future happenings. These predictive statements, however, are rarely put forth as absolute statements of performance but are instead expressed as scenarios of a likely occurrence, given the data processed. In short, these predictions point to behaviors that are statistically “likely” to occur in the future.

At this point it is worthy to mention that Predictive Analytics is a topic that is as vast as the mathematics that supports it. Many relevant models and approaches currently exist. I am going to discuss a small portion of this broad topic, and hope that those who wish to learn more can have at least have a launching point from which to do so.

Assessing risk is a staple of everyday business life, and the use of predictive analytics is prevalent in helping to mitigate or calculate the risk inherent in taking a specific business path, or in determining the probability that an event or behavior will occur in the future. The assessments that predictive analytics provide come in the form of values that correspond to the odds that an event or behavior is likely to occur in the future. There are three generally accepted categories of models used to perform predictive analysis: Predictive models, Descriptive models, and Decisive models.

Predictive models use historical data to assess how likely it is that a specific behavior will occur. These models are most often used to improve the effectiveness of an organization’s internal systems, as well as its marketing efforts. By matching patterns of likely behaviors, through the use of these models, with support systems, a new level of business efficiency is engendered. An organization can then apply this new found knowledge to perform risk analysis or produce a working knowledge of customer performance, fraud detection and loss prevention, it can detect organizational system “bottlenecks” and root out any subtle inefficiencies in manufacturing  processes, as well as in customer relationship management. Using a Predictive model, a business can take a snap-shot of risk in every business transaction, and in every customer relationship. The Predictive model is primarily a discreet model, which means that it uses “live” data to rank the likely behavior of one customer or demographic client group during one transaction. The end result in this model is a ranked order of likely behavior.

Descriptive models use the correlations that exist in “off-line” data to classify an organization’s income sources into viable groups. This model contrasts the Predictive model in that it seeks to identify many different relationships that may exist between income sources, like those relationships that exist between clients, and the organization’s products or services. This model does not rank likely behaviors or interactions, but rather is used to develop categories of likely participation. This type of modeling can be used to develop high level simulations, including those that employ a large number of individualized user agents who might be capable of performing highly specialized actions.

Decision models are used examine the relationships that exist between all of the variables surrounding any goal-oriented business decision. Primarily used in optimization, this model takes a data-driven approach to improving the logic that governs all goal-oriented decisions, and results in the production of business rules or guidelines to steer customer actions or circumstances. The Decision model uses historical data to do its work.

Customer retention is one of the key ingredients in business growth. By combining the above models and using the knowledge they provide, a business manager can judiciously steer his organization’s activities to meet corporate goals. Retaining customers, discovering new markets, creating targeted marketing, finding new opportunities to cross-sell, assessing risk, detecting fraudulent behavior, market prediction and implementing a high level of corporate efficiency to maximize profits are only a small portion of the benefits gained from actively using Predictive Analytics to read the proverbial tea leaves of any business organization.

Copyright 2008 - Dashboard Insight - All rights reserved.

Tweet article    Stumble article    Digg article    Buzz article    Delicious bookmark      Dashboard Insight RSS Feed
 
Other articles by this author

Discussion:

No comments have been posted yet.

Site Map | Contribute | Privacy Policy | Contact Us | Dashboard Insight © 2017