SAP Startup Focus Member Zementis is the Adobe Reader of Complex Data Science Models
Mark Rabkin is the Director of Business Development at Zementis, SAP Startup Focus member providing software for predictive analytics. Read the startup’s founding story and use case below.
Started in 2004, Zementis was founded by Michael Zeller with the intention of lowering the barrier of entry for complex algorithms built by data science teams in the IT production environment and beyond in the enterprise. The mission was to enable clients to take advantage of smart, real-time decisions in their day-to-day operations.
Traditionally, companies take up to six months to deploy their predictive analytic models. In this scenario, whenever a data science team finishes building the best model, they must also custom code it into the production environment of the IT engineering team. This lengthy process has no place in the big data era, where data is generated fast and changes rapidly.
Use Case Challenge
The financial services industry comprises a diverse mix of products and services, as well as an equally diverse set of business processes that support them. Banking can take the form of retail, commercial and investment, and each of these segments has its own distinct attributes. Insurance is an amalgam of different products and services, spanning a variety of risk profiles. Capital markets includes debt, equity, and specialized derivative products, which must be valued before they can be created and traded. High-frequency trading and quantitative portfolio management represent a financial art form that seeks to identify and exploit patterns that the broader market has not discerned.
While each of these financial industry segments is distinct, all have one thing in common: the need for analytics. Predictive analytics plays a critical role in sifting through the big data sets associated with each element of the financial services world. Zementis makes that work easier, by supporting the management of multiple models, facilitating complex scenario analysis and accelerating the deployment of models from data science teams to the business teams (or machines) making critical finance decisions. Whether the task at hand is fraud detection, insurance risk pricing, marketing optimization for retail products or quantitative trading strategies, Zementis can help.
In the case of fraud and risk management, Zementis can be applied to detect, prevent and reduce the occurrence of fraudulent activity. ADAPA® helps financial institutions to detect anomalous patterns and take proactive measures to stop abuse, minimizing risk. As a predictive tool, ADAPA® also helps organizations preempt future occurrences of fraud by identifying likely sources of fraud and allowing the institution to enact preventive measures before a risk actually materializes. With UPPI™ for big data scoring, financial institutions are able to establish connections between disparate data sources and detect fraud patterns that they had not previously identified. In these ways, Zementis helps financial institutions reduce their overall risk profile, reducing costs and helping financial services professionals stay focused on delivering value for their customers.
One of Zementis’ large banks first adopted ADAPA for its cross channel fraud detection efforts, honing in on anomalies in financial transfers that could be possible indicators of money laundering.
The company also piloted Zementis’ Universal PMML plug-in that acts as a platform for deploying predictive analytics throughout the organization. Analytics based decisions that typically required months of analysis took only days with Zementis, meaning greatly increased real-time decision making capabilities for the company.
Successful utilization of the solution relating to reduced time-to-value for fraud detection prompted the customer to permanently embed predictive analytics functions with Zementis’ enterprise database platform UPPI.
For the many companies that feel they are not ready to take on this business process, Mark outlined how money literally drops to a customer’s bottom line. Here’s how:
- The overhead expense piece of deployment between data science department and IT department is significantly reduced l by exporting the model to Zementis. For a large enterprise this process typically takes 6 months from model built to model deployed in the production environment.
- The data science team is 2x more productive because they are not spending time on IT compatibility and management.
- Companies get a taste of what is possible as Zementis is able to put together quick predictive analytics models
What 2015 Holds
In the upcoming year, Zementis will drive continued adoption of its solution with customers and will be guided by five factors outlined by Mark Rabkin, director of business development in 2015 in his article “5 Predictions for Predictive Analytics in 2015“.