IBM Safer Payments Hands-On Modeling Training (V6.3)

• Mandator Structure and its elements
• Sandbox Enviornment
• Modeling Concepts in Safer Payments
• Index for Profiling
• Profiling based on index with sequence
• Profiling based on index without sequence
• Profiling using Formula
• Ruleset/Rule Creation/ Rule Action
• Simulation Workflow
• Simulation: Data Selection and Sampling techniques
• Simulation: Attribute usage
• Simulation: Rule Analysis
• Simulation: Rule Performane
• Simulation: Rule Scoring
• Simulation: Rule optimization
• Inbuild Model Components: Rule Generation
• Inbuild Model Components: Random Forest
• Supported external Model Components: PMML
• Supported external Model Components: Python
• Collusion Algorithm

IBM Safer Payments users (Fraud Analysts, Fraud Investigators and optional: System Administrators), IBM Lab experts, and IBM Business Partners.

• Business knowledge
• Some familiarity with statistical models
• Understanding Safer Payments data inputs concepts

Mandator Structure and its elements 

• Understand Safer Payments structure  
• Answer the Hierarchy question 
• Prioritization Execution 

Sandbox Environment  

• Understand Champion/Challenger Concepts 
• How to promote a model to production 
• How to copy elements from other challengers 
• How to revert the model in production 

Modeling Concepts in Safer Payments 

• Understand at the high level the inbuilt techniques vs external model capabilities 
• Answer the question of what are the Inbuild Model Capabilities 
• Answer the question of what are supported external Model Capability to Safer Payments 

Index for Profiling 

• Understand the basis of profiling in Safer Payments 
• Create Index without sequence 
• Create Index with sequence 

Profiling based on index with sequence 

• Understand Profiling concepts based on index with sequence
• Build a counter 
• Build a precedent
• Build a pattern 

Profiling based on index without sequence 

• Understand Profiling concepts based on index without sequence 
• Build a calendar 
• Build an event 
• Build device identification 

Profiling using Formula 

• Ability to create a formula 

Ruleset/Rule Creation/ Rule Action 

• Understand Rulesets and Rules 
• Build a Ruleset 
• Build a Rule 
• Create a Rule Action 

Simulation Workflow  

• Understand The Simulation Environment and workflow 

Simulation: Data Selection and Sampling techniques

• Understand data selection for simulation 
• Select data selections for simulation  
• Sample the data selection 
• Run the simulation 

Simulation: Attribute usage 

• Understand the role of the attribute usage settings 
• Select attributes for simulation  
• Run the simulation  

Simulation: Rule Analysis 

• Understand the concept of the Rule Analysis 
• Create a Rule Analysis 
• Analyse a Rule Analysis 

Simulation: Rule Performance 

• Understand the concept of the Rule Performance
• Create a Rule Performance 
• Analyse Rule Performance 

Simulation: Rule Scoring 

• Understand the concept of the Rule Scoring 
• Create a Rule Scoring

Simulation: Rule optimization 

• Understand the concept of Rule Optimization  
• Create a Rule Optimization Report 

Inbuild Model Components: Rule Generation 

• Understand the concept behind Rule Generation 
• Understand the setting parameters 
• Use of verification data set and training data set 
• Use Interactive Mode for Rule Generation  
• Analyse the Rule Designer parameters 
• Use Fully automated Mode of Rule Generation 

Inbuild Model Components: Random Forest 

• Understand the concept of Internal Random Forest 
• Understand the setting parameters 
• Use of verification data set and training data set 
• Run and Analyse results 

Supported external Model Components: PMML

• Understand how to import a PMML model into IBM Safer Payments 
• Understand how to map inputs and outputs of the model 
• Understand how to use PMML for decisioning 

Supported external Model Components: Python 

• Understand how to import a python script into IBM Safer Payments 
• Use python for pre-processing rules Use python for formula  
• Use python for modelling 

Collusion Algorithm 

• Understand the concept of the Collusion Algorithm 
• Understand how to set up a collusion algorithm: manually and automatically 
• Create and simulate the Collusion  
• How to invoke the collusion algorithm