Data Mining

by Susan Cinnella-Firriolo, CPA – March 23, 2018
Data Mining

Innovations in technology are changing the world. In exchange for certain conveniences, we allow organizations to collect details about us. And all of those details are being stored in databases. However, most databases are unstructured, making it challenging for CPAs and others to work with. That’s where data mining comes in.

What is Data Mining?

Data mining allows people to find useful information in large databases. Simple definition? Yes. Simple process? No. Evolving from calculus, data mining is a mix of three mathematical specialties: statistics, artificial intelligence (AI) and machine learning (ML).   

Why is Data Mining Important to CPAs?

Data mining automates repetitive, time-consuming tasks. Despite the efficiencies data mining creates, it is underutilized in the accounting profession. However, CPAs who are Certified in Financial Forensics (CFF) and Accredited in Business Valuations (ABV) have wasted no time adopting the technology.

Pretend you’re hired to perform a foren­sic investigation for the Pharaohs in the Egyptian desert. Several bars of gold are missing from the pyramids. The Pharaohs believe workers are taking them. As you walk through the desert, you wonder if the sand holds the answers. You could manually sift through the sand looking for proof, but you realize this will take too much time, the Pharaohs will think you’re stupid and you will not get paid. You decide a better idea is to use an advanced metal detection system. After the system learns what information to search for, you put it to work in the sand. A few seconds later, the system returns its findings and you analyze the data. You discover the workers are in collusion with cattle herders. The herders distract the Pharaohs while workers steal the gold and hide it in the sand. After the cattle are corralled, the Pharaohs go back to the pyramids. The herders retrieve the gold and bring it back to the village. In this example, the sand is unstructured data, the findings are useful knowledge, the learning is machine learning and the system is data mining.

Cyberattacks, Fraud and Forensics

Data mining for cyberattacks is tricky because malware patterns are constantly changing. Predictive analysis is powerful, but it’s not enough to fight cybercrime. Consider the massive data breach at Equifax. Data mining exposed a vulnerability in an open platform used by the consumer credit bureau. A software update was issued by the developer which would have patched the weakness in no time. However, Equifax disregarded the update and, as a result, exposed the personal, sensitive information of over 145 million consumers.

Data mining for patterns or outliers can expose fraud. Without professional doubt, fraudulent transactions may be overlooked, dismissed or just avoided. In the WorldCom situation, WorldCom refused to give auditors access to the entire accounting system. So auditors used data mining to discover a fraudulent $1.7 billion journal entry.

Business Valuation: Science or Art? 

Business valuation is a specialty based on statistics and judgment. Forecasting (sometimes confused with guessing) is affected by so many variables, but it tends to be the most practical method of valuation.

Despite advanced data mining methods, predicting the future is challenging. Valuations rely on math which is limited by the quality of data. In 1997, many people thought a $6 million valuation of Amazon was pulled out of thin air. Nevertheless, Amazon’s IPO valued the then internet bookseller at $438 million mostly due to rapidly increasing revenue projections. Now, 20 years later, Amazon is valued at around $464 billion. Did early investors buy into the metrics or the idea, or a combination of both?

What's the Future?

Without a doubt, data mining is taking over activities traditionally performed by humans. But this doesn’t mean all jobs will be surrendered to machines. Did cashiers disappear after scanners were created? Data mining will become part of every business process, relieving CPAs of menial tasks and freeing them up to provide more value-added services. As this happens, the accomplishments and skills of highly specialized CPAs will be in more demand.

This article appeared in the March/April 2018 issue of New Jersey CPA magazine. Read the full issue.