The ability of big data algorithms to react quickly and comprehensively when financial or trade sanctions are imposed is “of the essence,” according to IT expert, Faisel Saeed, senior manager of Anti-Money Laundering Inference Analytics at the Bank of Montreal. Increasingly, the financial sector is turning to big data specialists to comply with these and other regulatory measures. By coincidence, the day we spoke with Saeed, tanks of unknown origin entered Ukraine, and days later, several nations, including Canada, announced sanctions against Russia and Ukraine further to those imposed in March 2014.

“Employees who are most comfortable working with big data have a certain mindset. For example, instead of merely describing customer behaviour, they want to anticipate it.”

“We [look] at which individuals or organizations are linked to residential or business addresses in the high-conflict area and block their transactions,” Saeed explained. “An added challenge is that UN aid missions and bona fide charities also operate in the same areas. We must be careful to separate the wheat from the chaff.” The key differentiator lies within big data analytics.

During the span of his career, Saeed has seen vast changes in data handling. In the past, paper forms had to be filled out, and “companies were judicious about collecting data because they had to pay people to key in the data,” he noted. When customers could click on an internet link, phase one of big data began, and information that could be collected cheaply proliferated.

Banks have, by and large, recognized the strategic need for big data, as summarized in the accompanying figure. “Financial institutions are developing enterprise-wide big data repositories,” Saeed explained. A broad net can be cast to support holistic customer relationships that include credit card transactions, lines of credit, and ordinary banking.

Saeed copes with rapid change not only from regulators but also from customers. The customer’s previous history is the best indicator of future behaviour—but not always. When something differs from the customer’s previous usage, he said, the bank asks for additional verification. “Customers usually appreciate the fact that the bank is looking out for them,” he added.

Regulatory changes, such as the sanctions mentioned above, may be mandated by the Financial Transactions and Reports Analysis Centre, Office of the Superintendent of Financial Institutions (“OSFI”) in Canada, by the U.S., or by the UN (with the latter two funnelling through OSFI).

As phase two of big data has gained momentum, business and technical characteristics are also changing. New techniques for big data analytics permit the exploration and discovery of valuable information with less awkward manipulation of data.

Initiated by business users and set up by their IT departments, traditional analytics focus on structured and repeatable analysis that configures the data to meet requirements. Examples of traditional analytics and reports include profitability analysis and quarterly sales reports. In contrast, Saeed explained, big data analytics use iterative and exploratory analysis. IT delivers a platform to enable creative discovery, allowing the bank’s line of business to continually explore what questions can be asked. Examples of big data analytics are maximum asset utilization, brand sentiment, and product strategy.

Exceptional technologies are required to efficiently process large quantities of data. Such technologies include crowdsourcing, natural language processing, visualization, massively parallel processing databases, search-based applications, and distributed databases. System performance, commodity infrastructure, and low cost are important requirements for providing real-time processing. How much data can be retained in memory at any one time affects the speed of working with it.

From a human resources perspective, employees who are most comfortable working with big data have a certain mindset. For example, instead of merely describing customer behaviour, they want to anticipate it. Saeed, who earned a PhD in IT from Oklahoma State University, was among the first generation to work with big data. He says the second generation is “savvier on tools and techniques,” and big banks are looking to hire these experts.

“The gap can be filled through hiring people with the desired formal education in data science or by retraining existing employees,” he said. Saeed compared the current hiring shortages to the IT revolution two decades ago.

The future

The third phase of big data is already upon us, and it uses images and videos as a source. Social media can also be leveraged to gauge market sentiment and find relationships between people based on shared connections and attributes such as schools attended, professional group memberships, and employment information.

There exist several open issues in the area of big data such as privacy and control. On the technical side, the raw data often needs to be cleaned, prepared, and made consistent with data from other sources before it can be used. This “data wrangling” stage can consume 50 to 80 percent of a data scientist’s time, according to expert estimates in an article by Steve Lohr in the August 17, 2014, New York Times.

And big data keeps getting bigger. The volume of data grows by leaps and bounds such that the majority of data that now exists has been created in just the past few years. In addition, the volume of business data worldwide across all companies is projected to double every 1.2 years, Saeed added. Consider that Facebook alone handles 50 billion photos from its user base.

“With the amount of information now accumulating,” Saeed concluded, “the key for big data analytics is to deliver results more rapidly and accurately.”