Written by Mark Palmer, Sr. VP Data Analytics, TIBCO

An ever-growing torrent of data is shaking up the activities and strategies of many organizations. Enabled by improvements in storage capacity, this is especially true in the financial sector which has historically strong reliance on data. Now, there is a desire for a better understanding of data and how to use it to improve business decision-making. But the desire is accompanied by new challenges in accessing quality information from increasingly disparate sources of information.

From working out the best way to extract and aggregate data from multiple sources, to leveraging artificial intelligence (AI) for risk management and improving customer knowledge, the ability to exploit data is becoming a differentiating factor for organizations in the financial sector. For too long, the data and insights from data science have been far too separate from the industry expertise held by business users. The next generation of analytics will have to decrease that gap and put data science into the hands of those who have the most industry knowledge.

Analytics within Financial Sector
The financial sector has always been very data-driven. Compliance, risk, and trading are functions that rely heavily on data analytics, management, reporting, modeling, and more. Data visualization gives organizations the advantages of more agile and speedy decision-making, thanks to improved activity management. Topic modeling, sound and sentiment analysis are bringing the ability to detect customer dissatisfaction in the tone of their voice and then trigger an action to make them more satisfied.

Whether insurance, banking, or finance, the financial sector has always been an avid consumer of data, statistics, and digital methodologies. But the most striking development in recent years has been the exponential growth in the volume, variety, and velocity of data.  Data has proven to be worth its weight in gold in terms of driving innovation, improving offerings, lowering costs, and enabling new disruptive business models.

What impact will AI have on current risk management practice?
Risk management is a major concern for the executive boards of banks and insurance companies. Their success in this area is typically measured with a rating, the top one being the coveted triple A. Artificial intelligence and machine learning are useful in many ways here. Statistical concepts like auto-encoding or linear regression help detect fraud while machine learning continuously improves the models and helps the user easily identify a higher likelihood of risk or fraud. AI offers powerful new tools for risk management by aggregating ever larger volumes of data and analyzing it faster than any human could.

AI does indeed offer powerful new tools for risk management by combining and analyzing large volumes of data. Data analytics tools bring the power of AI to industry experts — traders, risk managers, auditors, inspectors and others— providing them with a simple way to carry out complex analyses. An example would be using AI to aggregate millions of financial transactions based on the similarity of their risks. Industry experts can then easily use analytics tools to analyze the results, interpret the calculation parameters, and detect any abnormal concentrations of risks.

AI helps to solve the challenges of fraud detection, predict events like a customer running into debt, or predict the future behavior of a customer.  Fraud is keeping pace with developments in technology and fraudsters are always looking for new opportunities. Financial institutions are being forced to develop tools with increasingly powerful algorithms to identify anomalies, missing data, patterns that should not exist, etc. Catching fraud depends on an employee’s ability to analyze data and understand the situation in order to determine whether or not fraud has occurred. This requires tools with the ability to aggregate and analyze large volumes of data, combined with powerful visualization tools to understand the results quickly.

For example, data analytics help organizations create dashboards with real-time fraud detection, as well as alert analyses and decision-making functionality. The algorithms can be improved as experience is gained, thus creating a virtuous circle of learning. These cycles are often quite mature in relation to payment card fraud but are still in their infancy for other types of fraudulent activity.

The reduction of the fraud detection cycle time often involves adapting the infrastructure and the capacity to populate big data environments with real-time data. However, most improvements in fraud detection are studied in data labs, a typical scenario for any data science advancements. Unfortunately, with fraud as with other areas covered by data science, the implementation of various technologies and methods in data labs is unlikely to deliver an ROI since they result in use cases that are too far removed from the field and the business. This often makes them unsuitable for a wide-scale roll-out. In this context, data science platforms facilitate sharing and collaboration by uniting all the data tasks in a single tool and creating a link between what is developed in the data labs and how it’s used by business users.

What potential does analytics hold for increasing customer knowledge and improving the customization of financial offerings?
Banks and insurance companies are increasingly using scoring to compare financial products based on user profiles. Clustering formulas are also integrated in data science software for carrying out this type of comparison. The best way to improve the quality and traceability of financial data is to allow users – the industry experts – to get a hold of the data themselves. If only data scientists have access to the data, that puts an extra step between the industry expertise and the data and much crucial time and insight can be lost.

The potential for analytics in the hands of business users is huge and requires several stages, starting with a descriptive analysis. Analytics enable 360-degree visualization of customer data and the implementation of algorithms like the categorization of customers with much greater detail and accuracy.  The next step is predictive analytics, which gives an understanding of the dynamics of customer characteristics in order to predict their intentions, followed by prescriptive analytics to assist the decision-making of sales managers.

Machine learning methods allow financial institutions to greatly improve customer knowledge to the extent that they can develop highly customized marketing activities. In some sectors, an offer or a promotion can be targeted at a microsegment or even an individual.

About TIBCO

TIBCO Software unlocks the potential of real-time data for making faster, smarter decisions. Our Connected Intelligence Platform seamlessly connects any application or data source; intelligently unifies data for greater access, trust, and control; and confidently predicts outcomes in real time and at scale. Learn how solutions to our customers’ most critical business challenges are made possible by TIBCO at www.tibco.com.