Business Intelligence Best Practices - BI-BestPractices.com

Collaboration. Communication. Community.

 
 
 Printer-friendly
 E-mail to friend
  Comments
ADVERTISEMENT
BI Case Study: Tracking BI Success

by Stephen Swoyer
Wildlife conservation organization WildTrack is tapping a data analysis solution from SAS to track and monitor a constantly changing inventory.

Wildlife conservation organization WildTrack is tapping a data analysis solution from SAS to track and monitor a constantly changing inventory—of endangered African rhinos, that is. Even though WildTrack is a non-profit, its SAS solution has generated some pro-bono ROI for a group that needs it most—the indigenous workers of Namibia’s Waterberg Plateau region.

ZÖe Jewell isn’t like other workaday professionals. To be sure, when she heads off to the office each morning, she typically totes a laptop—but more often than not, she’s also got a GPS receiver, insect repellent, and fresh water. And although Jewell could conceivably be called an “inventory analyst,” she’s certainly not dealing with a “supply chain” in the usual sense of the term. As co-founder—along with her husband, Sky Alibhai—of the wildlife conservation organization WildTrack, Jewell’s home “office” spans a wide swathe of central, south, and western Africa, stretching from Tanzania’s remote Ruaha National Park to the arid regions of northwest Namibia. Needless to say, it’s not your typical nine-to-five factory job, but WildTrack isn’t so different from many small- and medium-sized businesses. It is, after all, managing a constantly shifting inventory—in this case, of endangered African rhinos—on a shoestring of an IT budget. The not-for-profit WildTrack—like thousands of for-profit businesses—has embraced BI, as a means both to cut costs and to empower more effective decision making.

WildTrack taps a data analysis solution from SAS Institute Inc. to track and monitor a constantly changing inventory—of rhinos, that is. What’s more, its SAS-based inventory management system is also being used—as an outsourced service—by several other conservation groups, including organizations struggling to protect the endangered Bengal tiger and the fabled Iberian Lynx. Even though WildTrack is a non-profit, its SAS solution has nevertheless generated some pro-bono ROI for the people who need it most—the indigenous workers of Namibia’s Waterberg Plateau region.

The “Business” Problem

Today, there are five extant species of rhinoceros, at least four of which are dangerously close to extinction. In Africa alone, two species of rhinoceros—the black rhino, which was once among the most numerous of all rhinos, and the white rhino, which comes in two flavors—have suffered incredible attrition.

Three decades ago, about 70,000 black rhino roamed sub-Saharan Africa; during the 1970’s, more than 90 percent of these animals were killed; today, about 3,600 remain. The northern white rhino—much sought after for the size of its horn—is all but extinct: Today, fewer than 30 northern white rhino survive, all of them in the Democratic Republic of the Congo. Elsewhere, the white rhino population in Africa’s southern regions is comparatively better off, with a total of more than 11,000 animals.

alt

In 1992, Jewell and Alibhai founded Rhinowatch, the precursor to WildTrack. As part of their research, the husband and wife duo investigated the effects of invasive monitoring techniques— e.g., tagging, radio collaring, and other methods—on a black rhino population in the Sinamatella Intensive Protection Zone of Hwange National Park, Zimbabwe.

“Our early research showed that there’s a negative impact on fertility, particularly for females, but all sorts of research has showed negative impact on behavior, on ecology, on structure, all of these kinds of things,” confirms Jewell.

What’s more, the duo surmised, such techniques ultimately contribute little to scientists’ knowledge of the social habits, migratory patterns, and breeding behaviors of rhinos. What helped seal the deal, of course, was the high cost of invasive monitoring: Not only must researchers purchase expensive radio collar gear, but they also have to finance a support network that includes helicopter pilots (to locate and track the animals), veterinarians (to tranquilize the animals once they’re found), and other specialists. Clearly, Jewell and Alibhai felt, a new approach was needed. “[Invasive techniques] were extremely expensive to implement, and relied on outside expertise, vets and pilots, and so when things did go wrong, it was just enormously expensive,” Jewell says.

Inventing a Technique That FITs

It was at Hwange that Jewell and Alibhai developed a revolutionary new monitoring method, called footprint identification technique (FIT), that has emerged a decade later as an alternative to invasive monitoring.

Instead of tracking and tranquilizing an animal in the field, preparatory to tagging it or attaching a radio collar, FIT describes a method in which human interaction with animals in the wild is minimized. In the FIT approach championed by Jewell and Alibhai, animal tracks are photographed and then analyzed—using statistical software from JMP Software, a subsidiary of SAS—to positively identify and track individual animals over a period of time. Call it fingerprinting for the hoof-and-paw set.

Jewell and Alibhai had both been impressed by the ability of indigenous trackers to follow and identify individual animals—just by studying their tracks. They developed FIT with this example in mind, using paper, pencil, and a ruler at first. “When we started out, we were actually tracing the footprints and measuring them,” she explains. But surely, they reasoned, if a human being could use a ruler and a pencil to identify the unique trackprints of an individual rhinoceros, computer software could do the same thing—but even more quickly and conclusively. Rhinoceroses are, after all, massive mammals, so finding clean, well-defined tracks wouldn’t be a problem.

The idea had enormous potential, Jewell says. “If you’re tracking an animal, if you’re looking at footprints, you’ll probably find a lot more about its behavior and its ecology. You’re finding out other interesting things as you go along,” she explains. “The other thing is that there’s a lot of emphasis now being placed on the role of indigenous people being custodians of their own wildlife. The idea is to give indigenous people a hand in directing their own wildlife conservation.”

Because FIT was to rely heavily on the contributions of native trackers, the research duo believed they’d have a chance to do just that.

SAS to the Rescue

Of course, they were most excited about the potential research benefits of the FIT technology. If they could develop a reliable means of tracking rhinos on the ground, they could also collect, analyze, and mine a lot more information about them. Over time, they reasoned, Rhinowatch could create a comprehensive database populated with information about the social, environmental, and breeding habits of individual rhinos.

When Jewell and Alibhai decided to take FIT mainstream this year, they thought a name change was in order, and so WildTrack was born.

Before they could begin, Jewell and Alibhai had to determine if their technology vision for FIT was even feasible. “We knew what we wanted to do—we just didn’t know how we could go about doing it,” she concedes. That’s when Alibhai, with his background in statistical number crunching, contacted JMP, a respected purveyor of statistical software for data analysis.

On the surface, JMP’s software product seemed ideally suited for the requirements Jewell and Alibhai had defined: It was designed to help analysts ferret out the relationships and outliers buried in their data, and—of particular relevance to the FIT technology Jewell and Alibhai hoped to develop—could also link statistics with graphics. What’s more, when Alibhai contacted JMP about the project, parent company SAS also took an interest. “Right at the outset, SAS helped us enormously in that respect,” confirms Jewell. “JMP in particular has been wonderful and SAS is wonderful in the flexibility it offers. I don’t have really any experience with [data analysis], but JMP has been idiot-proof for me.” SAS helped Jewell and Alibhai flesh out their FIT vision with a mix of custom and off-the-shelf software offerings. On the custom side, Nigel Law of SAS U.K. developed an application—NiSAS—that lets researchers key in a customized algorithm to define measurements for any input image. This is important, Jewell says, because it means the technology could theoretically be adapted to monitor any animal that leaves a footprint. Feed NiSAS the right algorithm and an appropriate footprint image, and it will generate measurements that JMP can in turn use to identify the animal.

In practice, the SAS-based FIT solution makes for a pretty straightforward technology proposition. Researchers or game scouts (who are tasked with fighting poachers) take digital photographs of rhino tracks when they’re in the field. At the same time, they’re careful to place landmarks—e.g., rulers or sections of measuring tape—at key anatomical reference points on the track. When the images are transferred to a PC, they’re optimized in Adobe Photoshop before being fed into NiSAS, which uses the custom input algorithm to create derived points from the landmark cues in the photos. From there, NiSAS generates 93 different measurements based on the distances and angles between these points. After that, JMP takes over.

Loading the Database

In the JMP graphical environment, the footprint library of each animal is represented as a circle; when circles overlap, researchers know the tracks came from the same animal. If there’s no overlap, they’re from different animals. It’s that simple. “We take digital cameras out into the bush and we go and look for tracks of these animals, and when we find them, we go and take a digital footprint of the track or a series of footprints along the track. Then we put those images into the laptop and take measurements from the footprints, which gives us what we call a geometric profile of the foot, and we then feed that data into JMP,” Jewell explains. “And JMP will basically compile a database of footprints.”

Because the SAS software was originally developed for use with black rhinos, Jewell and Alibhai didn’t know if it could also work with white rhinos, which are a completely different genus with a much larger footprint.

In 1999, however, they successfully used FIT to conduct a census of the white rhino population on the Otjiwa Game Ranch in Namibia. In about three weeks, they accurately determined the white rhino population at Otjiwa—a total of 26 animals. What’s more, Jewell and Alibhai tracked and cataloged the physical features—e.g. horn shape and size—of each individual rhinoceros. They fed this information into a database and were then able to map the ranges and distribution of all 26 animals. The upshot, says Jewell, is that wildlife managers could—for the first time ever—effectively manage Otjiwa’s rhino population.

Like many enterprise IT organizations, WildTrack is constantly gathering new data, thanks to the efforts of researchers in the field, native trackers, and game scouts on anti-poaching patrols. Over time, the organization hopes that historical analysis of this data will yield a wealth of valuable data.


“There’s a lot of essential ecological and behavioral data that you can get from this information. You can break out straight away the range of the animal, where it’s moving on a day-today basis,” says Jewell.

In this respect, she continues, FIT is a big improvement over radio collars that require researchers to actively track animals. “With this [approach], you can also look at who else the animal is interacting with. You can look at social groups, social dynamics,” she continues. “With females, you can see when they’re calving, you can chart the age of the calf. We’ve actually successfully been able to age tigers from their footprints, and we can also age rhinos. We can also work out which is left and right, and which is front and back.”

Overnight Sensation

Once word had spread of the Otjiwa success, Jewell and Alibhai were asked to adapt their SAS-based FIT solution to help track and monitor other species, starting with the endangered Bengal tiger. The rest is history.

In 2002, the duo received the Smithsonian Computerworld award for Environment, Energy, and Agriculture. And in 2004, they officially launched WildTrack as a means to coordinate with other conservation organizations around the world. Today, WildTrack supports several research projects using footprint identification to track and monitor the Sumatran rhino—which is believed to be the most endangered rhinoceros in the world—and the ultra-rare Iberian lynx, among others.

In the for-profit business world, BI tools are typically embraced as a means to help cut costs and empower more effective executive decision making. The same is also true in the case of non-profit WildTrack: The SAS-based FIT solution is much cheaper than alternative approaches and, Jewell says, yields richer, more valuable data. “In terms of actually using the technology, it’s an awful lot cheaper than any sort of invasive monitoring would be. It’s probably one of the most cost-effective monitoring technologies there is,” she says. “The cost is the computer, the digital camera, access to the Internet, and a GPS. There’s no aerial monitoring, no drugs, no radio collars. It really is more cost effective.”

WildTrack is a non-profit organization, but its conservation activities have enacted broader economic ramifications, especially in impoverished Namibia, where safari tourism is sometimes a more lucrative business than farming—when exotic or endangered animals are on hand, at least. The Otjiwa Game Ranch is one example, of course, but farmers in other regions of Africa—such as Kenya’s Laikipia Plateau—have also rehabilitated sizeable tracts of land to help encourage tourism. This, in turn, has fueled a demand for endangered animals—which benefits both conservation efforts and local ranchers.

Thanks to WildTrack’s success, for example, Otjiwa now derives most of its income from selling pairs of young breeding rhinos to other conservation areas. Jewell believes this success story can be replicated in other locales, with other species. “It’s not only helping [the indigenous people] play a more active role in wildlife conservation, but it’s also creating jobs. We need trackers, people with this expertise to help locate and track these animals in the first place,” she concludes. “So it’s also helping to revive that ancient art.”


Recent articles by Stephen Swoyer

Stephen Swoyer -

Stephen Swoyer is a technology writer based in Athens, Ga. You can contact Stephen via E-mail at swoyerse@percipient-analytics.com.