First Look – FICO Xpress Optimization 7.3

FICO Xpress Optimization Suite is part of the FICO Decision Management tools suite. It can be used in conjunction with various FICO applications too. It is used directly with their Marketing product and is embedded in their Decision Optimizer product that can generate optimal rules for their Originations, Customer Management, Collections and Fraud products. Outside of financial services FICO have used Xpress to optimize retail shelf positioning and layout, energy pricing and generation (one of Xpress’ largest markets) and more generally in scheduling, logistics and planning problems.

To tackle any optimization problem you need to model the problem as an optimization problem, solve it, explore the various alternatives before deploying and maintaining the model. Xpress therefore has various features:

• Modeling
  • Easy to learn and use with a full IDE for Mosel, Xpress’ language.
  • Distributed modeling and cloud enablement to allow models to be built and solved in a distributed environment
  • Strong data connections, a full set of APIs and pre-compilation are also provided
  • Optimization Engine
    • Linear Programming, Mixed Integer Programming, range of Quadratic Programming options, Successive Linear Programming (used to solve non-linear problems) and Constraint Programming.
    • High performance, multi-core, APIs
    • Insight
      • Scenario compares based on different data and parameters
      • Identify limiting factors/spare capacity, trade-offs, sensitivities
      • Visualization

Customers include P&G, the NFL, American Airlines, Budget, Avis, amazon.com and a number of OEMs.

The core Xpress tool contains an editor, debugger, profiler, progress graphics, visualization, wizards, Mosel extensions and deployment. Recent releases have improved performance (always a focus for optimization tools) and added support for cloud-based solving. Xpress Insight has a number of capabilities that can be mashed up into a compound user interface:

• Scenario Management
  Vary all input data, modify resource costs and constraints, identify binding constraints as well as spare capacity so analysts and business users can understand the tradeoffs and sensitivities.
• Visualization
  out of the box table and graph views that can be customized.
• Customization
  Present mathematical models in business terms, compute KPIs and allow flexible configuration using XML.

This kind of environment is important in optimization because the result of a model is often not readily explicable – the best answer is mathematically the best but the drivers for this are too complex to be readily understood. Being able to interact with the model, test the edges, try different scenarios and compare them all helps make business owners feel comfortable with the result – the optimization equivalent of impact analysis. Without this, business owners will not be active participants in optimization (just as impact analysis tools in business rules are essential for business user rule management). Once the best model and results are clear then optimal actions/assignments can be deployed, business rules derived from the model can be deployed or the mode can be deployed itself potentially with a “remote control” wrapper to make it easy to consume in an enterprise SOA environment. These Insight tools can be embedded in HTML applications and can interact with the engines using Javascript.

You can get more information on FICO Xpress here and FICO is one of the vendors in our Decision Management Systems Platform Technology report.