Name
WG 7 Session B - Data Quality & How Data Quality Issues Impact QRA
Date & Time
Tuesday, April 13, 2021, 2:45 PM - 4:15 PM
Description

Purpose

One of the categories by which data quality is assessed is pragmatic quality. It means the degree to which data is appropriate and useful for a particular purpose of data consumers. In particular, the data output from Risk Analysis should provide actionable results to pipeline operators and decision makers. 
In this Session we will discuss what actionable results mean, focussing on advantages of actionable results over merely descriptive results. Actionable analysis results enable efficient reasoning about the actual reality that underlies the data, as opposed to limited reasoning about the data itself. 

A deeper insight into the risk domain, stemming from actionable results, allows for a focussed identification and optimal prioritization of preventive and mitigating measures, based on specific overall objectives of a decision maker (e.g., to meet an overall risk target over time, or to determine dominant driving causes of failure). The reasoning based on actionable analysis results can proceed systematically from effects to causes, thus supporting human reasoning and problem-solving in complex risk domains. 

The type of Risk Analysis that produces actionable output must be based not only on input data, but—crucially—on a conceptual model in which known (or assumed) causal dependencies among the elements of the risk domain are quantitatively and rigorously encoded. The discussion in this Session will thus also include consideration of three general types of models that could be incorporated into risk analyses: deterministic, probabilistic, and causal models. 
 

Session Type
Working Group
Virtual Session Link
Add/Remove Session
 
 
Session Category
Working Group 7
Name
WG 7 Session B - Data Quality & How Data Quality Issues Impact QRA
Description

Purpose

One of the categories by which data quality is assessed is pragmatic quality. It means the degree to which data is appropriate and useful for a particular purpose of data consumers. In particular, the data output from Risk Analysis should provide actionable results to pipeline operators and decision makers. 
In this Session we will discuss what actionable results mean, focussing on advantages of actionable results over merely descriptive results. Actionable analysis results enable efficient reasoning about the actual reality that underlies the data, as opposed to limited reasoning about the data itself. 

A deeper insight into the risk domain, stemming from actionable results, allows for a focussed identification and optimal prioritization of preventive and mitigating measures, based on specific overall objectives of a decision maker (e.g., to meet an overall risk target over time, or to determine dominant driving causes of failure). The reasoning based on actionable analysis results can proceed systematically from effects to causes, thus supporting human reasoning and problem-solving in complex risk domains. 

The type of Risk Analysis that produces actionable output must be based not only on input data, but—crucially—on a conceptual model in which known (or assumed) causal dependencies among the elements of the risk domain are quantitatively and rigorously encoded. The discussion in this Session will thus also include consideration of three general types of models that could be incorporated into risk analyses: deterministic, probabilistic, and causal models.