From scenarios to dedisions

supporting decision making with exploratory and robust analysis

Research output: Book/ReportReportProfessional

Abstract

This report describes a case study that applies Robust Decision Making, RDM, a methodology that has recently been developed by RAND Corporation to support decision making under deep uncertainty. Robust decision is one that perform well over a wide range of future conditions. Deep uncertainty describes a common situation in which the parties to a decision don't know or cannot agree on probability distributions of key variables or models that relate the actions to consequences. When this situation prevails the traditional approach of striving for an agreement on assumptions as a first step of the decision making process may lead to a gridlock. That is why RDM uses agree-on-decisions approach. It inverts the classic decision making set-up: The analysis starts with existing plans or strategies and tests their performance in a multitude of futures covering all suggestions of the participating decision makers. The basic idea is that by using every participant's views in the analysis it may be possible to agree on a decision after carrying out the analysis even if the views on future conditions, model structure etc. differ. The approach is demonstrated by analysing nine alternative plans for the Finnish energy system. By giving a value range for the key variables a sample of 5000 plausible future conditions was defined. Every plan is tested in each of the future conditions. The most promising plans go through the vulnerability analysis. It is a statistical procedure to identify and present, in a concise way, those futures where the plans perform poorly. The discovery process not only identifies the factors but also reveals their value ranges which cause the poor performance. This range-information is essential in assessing the importance (likelihood) of the futures and in developing the plans to make them perform better in those specific circumstances, i.e., to make them more robust. Any model, especially existing ones, with decent running time can be used in the RDM process. In the RDM approach a model forms just one building block of the step-wise analysis process that uses a mixture of tools: the whole procedure is a modular one giving flexibility for carrying out the analysis focusing on points that are important for the problem at hand.
Original languageEnglish
PublisherTEKES
Number of pages63
Publication statusPublished - 2016
MoE publication typeD4 Published development or research report or study

Publication series

NameNeo-Carbon Energy Working paper
PublisherTEKES
Volume5

Fingerprint

Scenarios
Decision making
Uncertainty
Process analysis
Probability distribution
Robust decision making
Factors
Methodology
Energy systems
Decision-making process
Vulnerability
Decision maker

Keywords

  • simulation
  • robust decision making
  • exploratory modelling

Cite this

@book{1ce95e4044da41c5ace290ae9013a63e,
title = "From scenarios to dedisions: supporting decision making with exploratory and robust analysis",
abstract = "This report describes a case study that applies Robust Decision Making, RDM, a methodology that has recently been developed by RAND Corporation to support decision making under deep uncertainty. Robust decision is one that perform well over a wide range of future conditions. Deep uncertainty describes a common situation in which the parties to a decision don't know or cannot agree on probability distributions of key variables or models that relate the actions to consequences. When this situation prevails the traditional approach of striving for an agreement on assumptions as a first step of the decision making process may lead to a gridlock. That is why RDM uses agree-on-decisions approach. It inverts the classic decision making set-up: The analysis starts with existing plans or strategies and tests their performance in a multitude of futures covering all suggestions of the participating decision makers. The basic idea is that by using every participant's views in the analysis it may be possible to agree on a decision after carrying out the analysis even if the views on future conditions, model structure etc. differ. The approach is demonstrated by analysing nine alternative plans for the Finnish energy system. By giving a value range for the key variables a sample of 5000 plausible future conditions was defined. Every plan is tested in each of the future conditions. The most promising plans go through the vulnerability analysis. It is a statistical procedure to identify and present, in a concise way, those futures where the plans perform poorly. The discovery process not only identifies the factors but also reveals their value ranges which cause the poor performance. This range-information is essential in assessing the importance (likelihood) of the futures and in developing the plans to make them perform better in those specific circumstances, i.e., to make them more robust. Any model, especially existing ones, with decent running time can be used in the RDM process. In the RDM approach a model forms just one building block of the step-wise analysis process that uses a mixture of tools: the whole procedure is a modular one giving flexibility for carrying out the analysis focusing on points that are important for the problem at hand.",
keywords = "simulation, robust decision making, exploratory modelling",
author = "Juha Forsstr{\"o}m",
note = "Project code: 101283",
year = "2016",
language = "English",
series = "Neo-Carbon Energy Working paper",
publisher = "TEKES",
address = "Finland",

}

From scenarios to dedisions : supporting decision making with exploratory and robust analysis. / Forsström, Juha.

TEKES, 2016. 63 p. (Neo-Carbon Energy Working paper, Vol. 5).

Research output: Book/ReportReportProfessional

TY - BOOK

T1 - From scenarios to dedisions

T2 - supporting decision making with exploratory and robust analysis

AU - Forsström, Juha

N1 - Project code: 101283

PY - 2016

Y1 - 2016

N2 - This report describes a case study that applies Robust Decision Making, RDM, a methodology that has recently been developed by RAND Corporation to support decision making under deep uncertainty. Robust decision is one that perform well over a wide range of future conditions. Deep uncertainty describes a common situation in which the parties to a decision don't know or cannot agree on probability distributions of key variables or models that relate the actions to consequences. When this situation prevails the traditional approach of striving for an agreement on assumptions as a first step of the decision making process may lead to a gridlock. That is why RDM uses agree-on-decisions approach. It inverts the classic decision making set-up: The analysis starts with existing plans or strategies and tests their performance in a multitude of futures covering all suggestions of the participating decision makers. The basic idea is that by using every participant's views in the analysis it may be possible to agree on a decision after carrying out the analysis even if the views on future conditions, model structure etc. differ. The approach is demonstrated by analysing nine alternative plans for the Finnish energy system. By giving a value range for the key variables a sample of 5000 plausible future conditions was defined. Every plan is tested in each of the future conditions. The most promising plans go through the vulnerability analysis. It is a statistical procedure to identify and present, in a concise way, those futures where the plans perform poorly. The discovery process not only identifies the factors but also reveals their value ranges which cause the poor performance. This range-information is essential in assessing the importance (likelihood) of the futures and in developing the plans to make them perform better in those specific circumstances, i.e., to make them more robust. Any model, especially existing ones, with decent running time can be used in the RDM process. In the RDM approach a model forms just one building block of the step-wise analysis process that uses a mixture of tools: the whole procedure is a modular one giving flexibility for carrying out the analysis focusing on points that are important for the problem at hand.

AB - This report describes a case study that applies Robust Decision Making, RDM, a methodology that has recently been developed by RAND Corporation to support decision making under deep uncertainty. Robust decision is one that perform well over a wide range of future conditions. Deep uncertainty describes a common situation in which the parties to a decision don't know or cannot agree on probability distributions of key variables or models that relate the actions to consequences. When this situation prevails the traditional approach of striving for an agreement on assumptions as a first step of the decision making process may lead to a gridlock. That is why RDM uses agree-on-decisions approach. It inverts the classic decision making set-up: The analysis starts with existing plans or strategies and tests their performance in a multitude of futures covering all suggestions of the participating decision makers. The basic idea is that by using every participant's views in the analysis it may be possible to agree on a decision after carrying out the analysis even if the views on future conditions, model structure etc. differ. The approach is demonstrated by analysing nine alternative plans for the Finnish energy system. By giving a value range for the key variables a sample of 5000 plausible future conditions was defined. Every plan is tested in each of the future conditions. The most promising plans go through the vulnerability analysis. It is a statistical procedure to identify and present, in a concise way, those futures where the plans perform poorly. The discovery process not only identifies the factors but also reveals their value ranges which cause the poor performance. This range-information is essential in assessing the importance (likelihood) of the futures and in developing the plans to make them perform better in those specific circumstances, i.e., to make them more robust. Any model, especially existing ones, with decent running time can be used in the RDM process. In the RDM approach a model forms just one building block of the step-wise analysis process that uses a mixture of tools: the whole procedure is a modular one giving flexibility for carrying out the analysis focusing on points that are important for the problem at hand.

KW - simulation

KW - robust decision making

KW - exploratory modelling

M3 - Report

T3 - Neo-Carbon Energy Working paper

BT - From scenarios to dedisions

PB - TEKES

ER -