Print this page

Estimated reading time: 2 minutes, 13 seconds

Long Way to Go In Audit Data Analytics

Alan Anderson, RADAR InitiativeData analytics and data visualization promise a new way of conducting audits. But there is a lot of work to be done to develop the technology and prove the validity of these techniques in replacing traditional audit methodology. That was the message of a panel discussing the Rutgers AICPA Data Analytics Research (RADAR) Initiative recently.

The main promise of Radar is to "Move away from a sampling approach", according to Alan Anderson, executive director of the Radar Initiative. With current technology able to deal with big data, "We can deal with 100 percent of transactions," Anderson said. With the ability to analyze all transactions in real time, auditors will be able to avoid sampling error inherent in the current methods.

Radar was initiated a year ago by the Rutgers Business School and the American Institute of CPAs. The announced goal of testing new techniques and methodologies and encouraging research into audit data analytics.

The challenge is how to identify outliers and anomalies and then deal only with those that represent potential risks. Anderson said promoters must develop a sound methodology to identify "an exception that might need to be deeply looked at" versus those that require no further examination. Radar must also be able to prove the validity of the decisions and analytics make in identifying and categorizing exceptions and filtering them from the mass of data.

"The theories are fantastic," Anderson said. But he continued, it must be demonstrated that "these filtering techniques do produce acceptable results." The main need right now is for Radar to acquire large amounts of data to test. "We are also looking at various type of regression analysis," Anderson said.

Another component of the new approach to auditing is process mining. "It's in its infancy," according to Brian Miller, national assurance partner with BDO. But the goal is process mining "tells you how many times something deviates from process," he continued. "You potentially have the ability to automate part of the control risk assessment."

Process mining will be able to identify behavior that may occur. Process mining will tell you if it actually happened," he said. The system needs to be to distinguish between actions that were deliberately bad and those in which "a person was honest and didn't know they could do these things."

Data visualization is similarly in its early stages, according to Reid Simon, a manager with Crowe Horwath. Visualization, he said, means going beyond pie charts and being able to represent data graphically. However, standards have not yet been developed.

"What we are hoping to accomplish with this is identify approaches for being able to visualize audit data and incorporate visualization in the audit process," he said.

Read 5004 times
Rate this item
(0 votes)