Newsletter November 2005
 |
Email content |
 News
|
Vose Consulting partners with Decisioneering
Conferences, seminars and presentations
|
Upcoming Risk Analysis Courses |
Introduction to Quantitative Risk Analysis
Inverness, Scotland,
28-30 Nov 2005
Risk Analysis for Managers
London, UK,
14 Jan 2006
Corporate Risk Analysis
London, UK,
15-17 Jan 2006
Animal Health Risk Analysis and the role of Epidemiology
Pre-SVEPM, Exeter, UK, 26-28 March 2006
|
Risk Analysis tip |
"Modeling correlations"
|
|
News
Vose Consulting partners with Decisioneering to provide clients
with risk analysis expertise
Vose Consulting announces that it has partnered with Decisioneering, Inc., the Denver based publisher of
Crystal Ball ® risk analysis software, to launch a complete risk analysis consultancy service. Services include software development, risk analysis training, modeling and auditing, and developing strategies for dealing with risk... Read more
top
Conferences, seminars and presentations
-
As part of the executive MBA-class 'Risk and Decision Analysis', at the University of Texas, Dallas,
Vose Consulting partner Dr. H. Groenendaal gave a guest-lecture on "How to use expert opinion in risk analysis", which covered how to use expert opinion most effectively in risk analysis and ways to prevent common pitfalls.
-
At the Crystal Ball Oil & Gas User Group, Dr. H. Groenendaal presented "Distributions in Oil & Gas - How to Make Your Choice Not Random". The presentation covered the various probability distributions available in Crystal Ball and gave the attendees guidelines on which distributions to use in Oil & Gas risk modeling to make more accurate models.
-
On November 31st, 2005, senior partner David Vose will give a key-note speech at a
colloquium
of the European Food Safety Agency on risk, titled "Methodologies and challenges in animal diseases and food-borne biological risks"... Read more
|
Upcoming Courses
Introduction to Quantitative Risk Analysis, November 28th - 30th
Vose Consulting is offering this 3-day quantitative risk analysis course in Inverness, Scotland. There are still a few places left: click here if you are interested in attending this course or would like to have more information. For reservations and any further information contact us on  .
top
Risk analysis for managers, 14th January 2006
This short course offers busy managers the opportunity to learn when risk analysis can be useful to them, how to plan an analysis from devising the right questions to allocating the right resources and how to use the results to make better decisions.
The course will be held the day before the 3-day Corporate Risk Analysis course in
London, UK
(see below). Designed for managers, the 1-day course for managers is also of great value to analysts attending the corporate risk analysis course to better understand what their managers need from them. For reservations and further information contact us on .
top
Corporate Risk Analysis, 15-17th January 2006
Vose Consulting is offering their highly popular business risk analysis 3-day course in
London, UK. Check out the course description here for more information. For reservations, contact us on .
top
Animal Health Risk Analysis and the role of Epidemiology, 26-28th March 2006
This 3-day course is designed for anyone working in animal health or veterinary epidemiology who needs to conduct, present or critique risk analyses and needs to understand the relationship between Epidemiology and risk analysis. For more information see here or email .
top |
Risk Analysis Tip (a sample ModelAssist topic)
Modeling correlations
The following Risk Analysis Tip has been drawn from material in ModelAssistT. ModelAssist users can consult the ModelAssist-references (in the form of Mxxx) for additional information. To read more About ModelAssist and get a free download of the demo version, click here.
Introduction
A mistake often made in risk analysis models is not to take in to account relationships between input variables. This can cause large under- or over-estimates of the risks involved. This risk analysis tip describes several methods of including correlations in your model.
Not just one method!
In working together with clients, we often find that when people talk about (and model) correlations, they only consider using the rank-order correlation method. This is however not the only method available for modeling relationships between variables. There are, in fact, quite a few methods available, all with certain assumptions, advantages and disadvantages.
1. Rank order correlation
This method is quick and easy because both Crystal Ball and @RISK have build-in features to include rank-order correlations. Furthermore, rank order correlation does not need to model the direction of influence, so one does not have to specify which variable is dependent on which.
One of the disadvantages of this method is that it is not very intuitive and it is difficult to select the appropriate correlation coefficient. If one is simply seeking to reproduce a correlation that has been observed in previous data, the correlation coefficient can be calculated directly from the data (to see how, see ModelAssist, reference M0389) (Excel does not have a simple function for this, although the function CORREL( ) is usually a good approximation for roughly symmetric distributions). The difficulty appears when attempting to model an expert's opinion of the degree of correlation between distributions. A rank order correlation lacks intuitive appeal and it is therefore very difficult for the expert to decide which level of correlation best represents her opinion.
2. Copulas
Copulas are multivariate random variables with Uniform(0,1) marginal distributions. Almost all probability distributions can be generated using an inversion method which requires supplying Uniform(0,1) random variables. Thus, copulas can be used to generate correlated random variables. The method is similar to rank order correlation but there are a considerable number of different copulas producing different correlation patterns, allowing for far greater control of correlation. The copula parameters can also be estimated and compared statistically, unlike rank order correlation. Copulas have become increasingly popular in many fields, including finance, human health and engineering risk applications. Vose Consulting has developed its own in-house suite of copula to ols to use in its consulting projects giving clients better risk analysis results.
3. Envelope Method
This method has the dependent variable being modeled by a distribution whose parameters are functions of the independent variable. It is well suited to modeling expert opinion of correlated variables, and is easy to use and check. It can model one-to-many relationships, but is difficult to adapt to many-to-many relationships and would require determining a logical sequence of relationships.
4. Using Lookup Tables
This method modifies a distribution or selects from different distributions to model a variable, according to the value that is generated for a variable it is being influenced by. The lookup table method is well suited to modeling expert opinion of correlated relationships, and can model one-to-many influences, but is difficult to adapt to many-to-many influences and would require a sequence of influence.
5. Conditional Logic
There are various functions (e.g. IF(), AND(), OR()) in Excel that allow one to build up a logic that makes a Cell switch between values according to other Cell values. We can capitalize on these features to build up relationships between our model variables.
More details and models.
ModelAssist provides you with a much more detailed explanation (including example spreadsheets) of the above methods.
ModelAssist
- The material within this 'Risk Analysis Tip' comes from one of over 500 risk analysis topics available in ModelAssist, which gives a more detailed explanation of the above methods and any risk analysis techniques involved.
> Learn more about ModelAssist
top |
|
