The Gestalt of Data

Neugents, visualization analyze data and predict behavior

They come in droves. They come by car, plane, train, boat and helicopter. Many times, they come by the busload, emptying into the lobby and lurching down the aisles of the casino floor until they have tugged and pulled on every slot machine they can find.

They are slot fanatics, and keeping track of the mass of humanity that pours into a casino on a daily basis — and gathering accurate details of their activities — would seem a very difficult task. Even if you could do it, what possible sense could one make from the data?

Its a dilemma faced by technicians at Harrahs Entertainment, which runs 21 casinos in the U.S. with a combined database of 19 million players. While it supplies plenty of card tables for the high rollers, Harrahs also caters to the slot machine enthusiast. And since those machines are tied to a single database, Harrahs spokesman Gary Thompson says, it makes sense to draw as much useful data from it as possible.

Harrahs recently began installing Compudigm Internationals data visualization technology at its Las Vegas headquarters, with plans to expand its use to more casinos.

Data visualization has begun to move away from "fringe technology" to the mainstream of business intelligence. Where BI can aggregate data so its easily recognized and usable in real-world applications, data visualization gives the aesthetic interface to that data, and provides a view that may draw attention to details that might otherwise be missed in text readouts.

Intelligence and Then Some

"We call it aided intelligence, because we have an environment thats able to represent data in a way that businesspeople can see in such depth and worth in real-time and effect performance," says Andrew Cardno, president and CEO of Compudigm. The genesis for Compudigm was the gaming industry, but since then the company has extended its technology to financial services and the telecom industry.

The visualization software will allow Harrahs to view the flow of traffic across the casino floor. "Slots make up the bulk of Harrahs revenue," Thompson says. "It shows us which slot machines are popular with our customers and which ones are profitable for us, so we know which slots to keep or install more of."

Its no mystery the layout of every casino is a strategic enterprise, meant to keep players gambling as long as possible. "Theres an art in how you place the machines around the casino and we have people who are experts on that," Thompson says. "The purpose of data visualization is to enable us to see at a glance how we should be organizing the casino."

With the software, Harrahs can get an emulated picture of the casino floor with details on each machine, such as what type of machine it is, how often it is played, how much money it brings in and how much it coughs up. Harrahs can also tie the software into its rewards program, which acts as a frequent-flyer program for gamblers, meant to encourage loyalty to Harrahs casinos nationwide. Harrahs can market specific products to individual customers, based on their profiles.

At Ford Motor, one of the most important reports produced each year is the 126 Report, a summary of each Ford dealerships warranty claims. The huge report is packed with valuable information — so much so, says Michael Jansen, Fords North American Warranty Systems design manager, that dealers have trouble pulling out the nuggets of information that are most pertinent to them.

Ford worked with New York-based BI software vendor Information Builders to create a Web-based version of the report that dealers can access and analyze using a variety of visual tools.

Jansen says the visualization tools mean that the big picture is immediately clear. If a dealerships warranty record is out of line, for example, or even starting to go out of line, one or more elements of the corresponding graph will show up as yellow or red, indicating that the dealer has exceeded acceptable limits.

Another surprise benefit of the Ford implementation is its ability to move information up and down the food chain. Mechanics at some dealerships who previously had no access to the statistics in the 126 Report are now receiving graphical charts with vital statistics. For example, if they see the warranty trend for brake repairs is edging above a certain threshold, they know they need to focus on the work being done by the brake repair team before it becomes a financial problem for the dealer.

Visualization is only one innovative technology changing the face of BI today. Another is neural network agents, which work a little more like artificial intelligence (AI) than BI.

Neugents were developed by researchers at Case Western Reserve University in Ohio in the early 90s, and then a company called AI Ware was built to harness the technology. While AI Wares technology was state-of-the-art, the company didnt have the funds or market presence to do anything with it until AI Ware was acquired by Computer Associates International, which has since been integrating the technology — which it calls Neugents — throughout its product line.

Saving Lives

New Scotland Yard, headquarters of Londons Metropolitan Police, uses Neugents to predict crime behavior. Neugents analyze data about criminals in a certain area and whether theyre more likely to commit a certain crime, on a certain day, at a certain time of day, in certain kind of weather.

"Their business isnt saving dollars and cents, but saving lives," says Carl Hartman, vice president of business development at CA. "The idea is a criminal hears a knock on the door and theres a cop that asks if hes planning on going out tonight. The attention could prevent the crime from happening."

Preventing Outages

There are more uses of the technology in the commercial world. Most important for e-business managers is maintaining network uptime. After most major site outages, theres a trail of data a mile long that, if known beforehand, would have alerted the managers that an outage was about to occur. Neugents can constantly comb through network traffic and send out alerts when a network outage is close at hand.

"If you can predict when thats going to happen and get an alert that the networks going to fail in 17 minutes, that gives you 16 minutes and 59 seconds to prevent it from happening," Hartman explains. "We consider that business intelligence thats a critical factor for success."

For a major e-commerce site, preventing a network outage that could cause millions of dollars in lost business and lost customer loyalty is a pretty easy sell.

Predictive analysis isnt a point-and-click solution, however. The accuracy is limited to the specific application and the data available to analyze, says Robert Blumstein, director of analytics research at IDC.

"I cant emphasize enough the quality of data on which [predictive analysis] is based," Blumstein says. "I think, today, organizations are working to make clean data a higher priority than they had in the past."

To make predictions based on data, Blumstein says, you first need data that works. But even then the data may only get you so far. For example, a car dealership can market minivans to an individual if it knows he or she has a family of four. However, if all the dealership knows is that the customer is a teacher, then pinpointing an effective marketing campaign for that individual would be difficult.

But even with clean data, Neugents, like data visualization, arent a replacement for a good BI implementation; rather, they are an augmentation to improve the intelligence at hand.

"Business intelligence is good, but its kind of backwards-looking," CAs Hartman says. "We find patterns in the data and can tell you whats going to happen next."