eXplainable Predictive Decisioning
summary
A standards-based pipeline combining PMML for predictive models and DMN decision tables is presented, where PMML models persist and are invoked from DMN so thresholds, scores and policy rules run in series to trigger automated actions. A banking fraud/risk example runs on Red Hat Process Automation Manager deployed to OpenShift with Prometheus monitoring, supports scenario simulation and regression testing, and allows swapping models without changing decision logic to produce explainable risk scores.
AI-written summary · about
chapters (8)
- 0:00Introduction
- 5:34ML building blocks
- 13:10Demo: customer automation
- 19:21Decision logic & risk
- 25:13Predictive models & rating
- 31:05Stakeholder dashboards
- 35:05Platform integration
- 37:10Q&A
related talks
15mOperationalizing Predictive Analytics in Education
53mIntroducing AI to the Flowable Process Engines
22mMachine Learning–Driven Third-Party Risk Management at Scale
34mSynergizing model and human decision-making using augmented machine learning
34mOpen Standards For Machine Learning Model Deployment
32mMaking Models Work: How an optimized ML lifecycle can drive better ROI
19mModel Performance Monitoring - A Practitioner's Perspective
50mPredictive Testing