The perfect pairing: data science and human expertise enhance our insight

If you’ve ever typed “are robots replacing humans?” into a search engine, you’re definitely not alone. However, while we harness the power of technology at Allianz Trade, we recognize that it’s the expertise of our human team members that brings the real added value. We believe that using the two in tandem creates more efficient processes that can help predict and, crucially, manage risk.

Effectively leveraging data science starts with understanding how it complements human analysis. As such, our team of data scientists works with other functions to identify how tech can improve processes.

Through this collaboration, we’ve developed a variety of machine learning models that support our teams across the company:

• Our underwriting model automatically assesses and greenlights certain small credit limit requests for clients – those that are simple, but time-consuming. This enables our employees to handle more complicated tasks that require human expertise and experience.
• Our fraud detection model highlights suspicious behavior before it becomes a problem. It considers multiple financial and non-financial factors
• Our non-payment risk model contributes to determining a buyer’s grade in our system. It pulls a huge amount of information  which informs our teams’ decision making.

The true common denominator in all our models is human input – every model we create is crafted using the methodology and concepts of our expert team.

When building new models, we start with a visibility assessment to determine if the idea is feasible and if we have sufficient data. By “sufficient data,” I mean enough information to find patterns, thereby optimizing the output. We call this “training” the machine, and it requires a lot of high-quality, labeled data. If we don’t have enough, it’s all hands on deck to input it manually.

We then perform a comprehensive interview with the team to get a deep understanding of their problem, their way of working and how the model can support them. All of our models are country-specific, so we rely on our colleagues to supply us with their expert input regarding day-to-day issues they encounter, as well as local regulations and contexts.

Once we have a full picture, we start building. Finally, the working model is integrated into the business’s information system, becoming a seamless component of our operations.

What we’re doing here at Allianz Trade isn’t magic. It’s data, models, statistics – but it is special. We’re always striving to stay ahead of the curve to provide the best possible service to our clients. Integrating machine learning into our entire value chain, and leveraging the expertise of our teams, enables us to do just that.