Analytics: Predicting the future (and past and present)
- By Rutrell Yasin
- Mar 13, 2013
Predictive analytics can help government agencies improve health care and quality of life, cut waste and abuse and even aid in the prediction of crime.
But law enforcement officials might not get the results achieved by the specialized police department in the movie “Minority Report,” which apprehended criminals based on the foreknowledge of psychics called “precogs.” Instead, police departments from Baltimore to Miami are using analytics software to forecast patterns and identify criminal hotspots in an effort to head-off crime and better allocate resources.
Predictive analytics is an area of statistical analysis that deals with extracting information from data and using it to predict trends and behavior patterns. The term inherently suggests the future, but predictive analytics can be applied to any type of unknown situation, whether it is the past, present or future, expert say.
“There are many disciplines within the analytic domain, and predictive analytics is one of them,” said Mark Cleverley, public safety director for IBM. Predictive analytics allows analysts to construct a model of the world as it might be. The objective is to determine with a degree of confidence that a situation might happen.
Cleverley noted predictive analytics is not a new discipline; it’s been applied by engineers and scientists for years. The military was an early adopter, using analytics to determine the likelihood of engine failures in vehicles. Another well-known application is the use of a customer’s credit history to determine his likelihood to make future credit payments on time.
Now, officials in federal, state and local agencies want to use predictive analytics to help make informed decisions based on increasingly large volumes and variety of data.
For instance, Miami-Dade County is collecting and analyzing data from across 35 municipalities to help them make smarter decisions about managing water resources, reducing traffic jams and fighting crime. IBM’s Intelligent Operations Center (IOC), styled as an executive dashboard, will let city and county leaders scoop up information from one department and share it with others, thereby improving access to pertinent and time-sensitive information, Miami-Dade officials said.
Miami-Dade County’s Parks, Recreation and Open Spaces Department is also using predictive analytics and smart metering to help monitor water consumption and identify leaks remotely, according to Carmen Suarez, Miami-Dade’s IT director.
The county has 263 parks, spanning 12,845 acres of land, the third largest in the U.S. In the past, the system’s aging water infrastructure had to be manually inspected to detect leaks or other problems. The county was also hit with rising water costs that drained more than $4 million annually from its coffers.
On top of that, the department had a complicated, labor-intensive process for collecting and analyzing historical data, all of which prompted officials three years ago to take a look at big data and analytics for managing city services.
IBM’s analytics dashboard will let county employees remotely monitor water consumption and detect leaks. The platform will also include a Web portal so they can easily view and share water consumption data with other departments and monitor and manage overall water usage. The department is expected to reduce water consumption by 20 percent and generate $860,000 in savings per year.
Big data on the beat
IBM predictive analysis tools will help the Miami-Dade incorporate intelligent policing techniques, reducing the time it takes officers to identify leads, investigate crimes and solve cases. The police department already had an advanced crime data warehouse, Suarez said. Using IBM’s technology, the police can link directly into park systems and other departments’ operations to collaborate with other agencies and organizations.
MDPD is using IBM SPSS (originally for Statistical Package for Social Science) predictive analytics tools to identify unique relationships and spot patterns. SPSS can hold information about stolen property, time of day, weapons used in crimes and victim details. From that information, analysts can create a model of the type of suspect who might commit a particular crime, and then generate and filter a suspect list to help analyze the cases faster — and in some cases, anticipate and prevent future events.
While Miami-Dade focuses on more efficient services to citizens, the Inspector General of Illinois’ Department of Healthcare and Family Services is using analytics to tackle insurance claims fraud. The department has used SAS Analytics for the state’s Medicaid program to identify and prevent overpayments to health care providers, said Wayne Thompson, SAS product manager for predictive analytics.
The SAS-based fraud platform uses historical data on previous fraud and abuse cases to develop fraud predictors. By using the insights from known fraud cases, the system can spot provider collusion and identify undiscovered fraudulent providers and criminal networks, avoiding significant fraud-related financial losses each year, officials said.
In the past, when agency officials suspected a person or provider of fraud, analysts had to perform a lengthy assessment, according to Weishin Wang, assistant bureau chief and project manager with the department. The agency has now fine-tuned its predictive models so the analytic software can direct analysts to targeted providers. It has also identified routines, with interrelated patient information, to spot suspicious networking activities among providers.
The agency can detect fraudulent activities such as time-dependable billing, non-corresponding medical claims and double billing. Thompson said the agency also is finding patterns in social networks through a process call dynamic networks association.
Agencies dealing with tax claims are increasingly turning to predictive analytics, including the IRS and the Australian Taxation Authority, which is using SAS Enterprise Miner for debt collection, Thompson said.
The IRS is also looking to analyze tax code changes and legislation as well as predict the impact of events such as Hurricane Katrina on tax revenues. For analysts to make decisions quickly, the analytics has to be in the database. At the IRS, for example, SAS Analytics is embedded in EMC’s Greenplum database, which performs massively parallel processing.
Performing analytics where the data resides is also a strategy Oracle has adopted. The company provides a predictive analytical engine called Oracle Data Mining within the company’s relational databases. “We want to have the right tool where the data lives,” said Peter Doolan, the company’s public sector CTO.
But the real power of big data analytics will be unlocked when analytic tools are in the hands of everybody, not just among data scientists who will tell people how to use it, according to Gus Hunt, the CIA’s CTO, during a recent seminar on Big Data in Washington, D.C.
“We are going to have to get analytics and visualization [tools] that are so dead-simple easy to use, anybody can take advantage of them, anybody can use them,” Hunt said.