IBM Safer Payments is an innovative real-time payment fraud prevention and detection solution for all cashless payment types. IBM Safer Payments provides not only model capabilities based on inbuilt tools, but also the option to import externally built fraud models for real-time decisioning.

In this course, all of the IBM Safer Payments model capabilities are presented in detail. The following modelling concepts are covered: index, profiling techniques (with and without index sequence), model components comprised of rulesets, PMML, Python and Internal Random Forest, elements of the simulation environment including Rule Generation and Internal Random Forest, as well as the sampling techniques. All these concepts will be followed by the hands-on exercises.

The objective of this course is to teach students the following topics:

• Mandator structure
• Modeling concepts
• Profiling concept
• Rulesets and Rules
• Simulation environment
• Analysis
• Model components
• Python callouts

• External: Fraud Analysts, Application and System Admins managing Safer Payments (optional)
• Internal: IBM Lab Services, IBM Support, IBM Technical Pre-Sales and IBM Business Partners

• Must be familiar with Unix command line navigation and configuration actions
• Some familiarity with statistical models
• Knowledge in Fraud Prevention for cashless payments

Day 1:

• Safer Payments Data Dictionary
• Modeling Approach (Internal & External Modeling)
• Examine Indexes with and without sequences
• Profiling in Safer Payments using index with sequence (Counter, Precedents, Pattern)
• Profiling in Safer Payments using index without sequence (Calendar, Events, Device Identification, Formulas)
• Introduction to Rules

Day 2:

• Introduction to Simulation workflow
• Sampling Techniques
• Rule Analyses and Rule Performance
• Rule Performance and Rule Scoring

Day 3:

• Internal modeling capabilities (Rule Generator and Random Forest)
• Exporting and importing data for external modeling
• Python callouts
• PMML Model Import
• Point of Compromise