Decision Making in the Abscence of Ground Truth
summary
Claims a repeatable method to make quantitative risk estimates without historical data using subject matter experts, producing management-ready KPIs and prioritization. Uses the Sheffield Elicitation Framework to elicit lower/upper bounds, median and quartiles from experts, projects those into numeric samples (e.g. generate 10,000 samples or smaller ensembles), then computes expected loss and ranks items — showing a functioning demo that returns a top 10 riskiest assets list and drillable tables for targeting maintenance.
AI-written summary · about
chapters (13)
- 0:00Start
- 0:48Agenda
- 2:32Introduction
- 13:41The Path to Success
- 15:31Elicitation Framework
- 19:45Challenging your value for L
- 21:43Median M
- 24:35Challenging your median
- 25:50Refining your median
- 27:54Placement of Q1 - examples
- 30:29The next step
- 31:25Consensus
- 33:36What would RIO think?
related talks
35mWhy Risk Management Requires Taking Risks: A Practical Guide...
50mQuantifying Swiss Cheese, the Bayesian Way
49mGetting to True Predictive Risk: Will Data Accuracy Thwart AI’s Potential?
29mEnhancing Quality Control Efficiency: A Dynamic Risk Threshold Approach
34mPrediction intervals
27mMachine Learning Practical Example
36mA Data Science Approach to Systemic Risk
25mOpening Remarks