28 Apr

The Science and “Magic” of Big Data

By: Alan Krumholz, Principal Data Scientist

 

How data science is changing the face of payments

The field of data science is growing rapidly, with an explosive demand across all industries for the acquisition and analysis of “big data.” Data science has the ability to take existing business practices to new heights and help make the best decisions possible. In the payments industry, where the ecosystem is constantly changing, the role data science plays is extremely valuable – especially regarding fraud management, merchant due diligence, and customer insight.

 

Data science is the combination of math, technology, and business knowledge, all working in synergy to extract knowledge from data. Each of these pillars needs to be strong in order to use data sciences effectively to help answer the following four questions:

 

Data-Science-Analytics-Infographic 20150427

 

  1. Descriptive: What happened?
  2. Diagnostic: Why did it happen?
  3. Predictive: What will happen?
  4. Prescriptive: What should I do about it?

 

These are the stages of the analytics spectrum as defined by Gartner. The first two, descriptive and diagnostic, focus on the past, while the latter two are forward-looking. In any industry, understanding what happened in the past, anticipating what will happen in the future, and knowing what to do about it leads to better business decisions.

 

Both sides of the payments value chain are turning to data science for help – legitimate businesses and cybercriminals alike. For example, illegal online pharmacies that sell prescription medicine need to process credit card payments, and paying a low processing rate is essential for their success. Because of this, these fraudulent businesses are extremely wary of transactions that can drive their rates up, or even worse, draw too much attention and get them shut down. To avoid this, cybercriminals have developed their own predictive models to stop suspicious transactions.

 

To keep fraudsters at bay, it is important to help honest businesses succeed. G2 is pioneering the advanced application of data science in the payments industry to help business portfolios continue to thrive and grow. Here are a few examples:

 

Predictive Merchant Scoring

Using data that G2 has collected over the past decade, plus additional third-party information, data scientists have created the G2 Compass Score®, a scoring model that is 99% accurate at predicting if a merchant will commit a violation in the future. This score helps acquirers and payment service providers quickly board low-risk merchants and spend more time assessing merchants they define as high-risk, thereby making better decisions about who they add to their portfolios.

 

Merchant Content Monitoring

data One of the most effective techniques to find violations online is to monitor the text on websites, using keywords that are related to content violations. However, as the volume of websites increase and cybercriminals become smarter, this method has become less efficient. To tackle this, G2 data scientists enhanced Persistent Merchant Monitoring with a predictive model that instead of looking for specific keywords, looks at ALL of the words on a website and statistically comes up with the probability of the website having any content violations. G2 uses a machine-learning algorithm that learns hundreds of decision trees based on this data, meaning it will assign different probabilities to different word combinations. This new model makes the process 1,000% more precise, exponentially reducing the number of false positive examples that need to be reviewed by our analysts.

 

Marketplace Counterfeit Detection

G2 data scientists have also built models to find counterfeit products on marketplaces. Starting with data from only a handful of examples of counterfeits on the marketplace, they created a model that would score the products and send the ones with a higher probability of being counterfeit to G2 analysts for a closer look. As feedback from analysts increased, their model grew from 14% actual counterfeit products per report to its current 99%. In the process, G2 uncovered two major players selling thousands of counterfeit products unbeknownst to the marketplace.

 

Data science is changing the face of the payments industry, allowing for improved insight into merchant due diligence and compliance, and giving businesses the information necessary to predict the future. Companies throughout the payments value chain must learn to embrace data science to make the best decisions possible and stay ahead of the curve. With over 10 years in the industry, a decade’s worth of data, and experienced analysts providing constant feedback to our statistical models, G2 is constantly applying data science to solve the payment industry’s toughest problems.

 

Share this