Talk:Event tree analysis

comments

In the Introduction/Definition I miss the reference to risk assessment as an application for ETA. I think it is worth including it not only because this wiki is devoted to this topic but also because you have a complete section for it later on. In the Theory section I would recommend to define the concept loss end states. I wonder if the third bullet point of this section is correct. The second sentence of the second paragraph of Methodology might have a flaw: start at the left?. In this same paragraph when you refer to 1s I would include 1f as well and would recommend to see the attached graph. I don't quite understand the 4th step (bullet point) of this section: indentify intermediate events. I have looked for more information in one of your references C.A. Ericson (2005) and this point is clearer for me in the original paper: Identify the pivotal events. Perhaps, it might be worth changing it. Event tree Analysis I miss something at the beginning of the second sentence of this paragraph. In Advantages, I wonder if the fith bullet point is correct: ineffective or effective countermeasures? The 9th point says easy to lean perhaps it should say easy to learn. I wonder if it would be possible to include some names or references to complete the last point,i.e. some of the software available. Last but not least, I would recommend to include more references. It is always advisable for a wiki. By the way, I wonder if ETA is the same as Classifying tree analysis. I have learn about it in the article by Cox and Popken (2008) recommended in this course, and I think it is a very interesting tool in RA. Should it be the case, I would include a mention to this statistical method. You can find additional references in this paper and a mention to the rpart algorithm of the stastical software R. Perhaps, you can use it to improve your wiki. Good luck 570acm570acm (talk) 21:08, 9 November 2013 (UTC)[reply]

In the theory section, a probabilistic risk assessment use is stated; however in the methodologies and equation section there is a more deterministic approach described. Distinguishing between instances when a single point probability is used at each intervention point in a deterministic type assessment vs. a stochastic type assessment using probability distributions at each intervention point run with multiple iterations might be beneficial.

Perhaps examples of the use of varying complexity ETA’s could be added to the page.
How is uncertainty addressed when an ETA is conducted?

570ps (talk) 02:45, 11 November 2013 (UTC)[reply]

Comment reply

In the Introduction/Definition I miss the reference to risk assessment as an application for ETA. I think it is worth including it not only because this wiki is devoted to this topic but also because you have a complete section for it later on.

I think that I would agree

In the Theory section I would recommend to define the concept loss end states.

I think that with my current edits it is easier to understand but I agree with your point.

I wonder if the third bullet point of this section is correct.

a loss or end state can be defined by the user those bullet points are just common end states.

The second sentence of the second paragraph of Methodology might have a flaw: start at the left?.

good catch!

In this same paragraph when you refer to 1s I would include 1f as well and would recommend to see the attached graph.

I agree

I don't quite understand the 4th step (bullet point) of this section: indentify intermediate events. I have looked for more information in one of your references C.A. Ericson (2005) and this point is clearer for me in the original paper: Identify the pivotal events. Perhaps, it might be worth changing it.

Using pivotal events/points is a preference as is intermediate events/points, I think that some people may understand it better one way or the other but I do see where you are coming from and I think that my added example may help the reader understand “intermediate events”. Let me know what you think about the change.

Event tree Analysis I miss something at the beginning of the second sentence of this paragraph.

I agree that was a little difficult to read

In Advantages, I wonder if the fifth bullet point is correct: ineffective or effective countermeasures?

Yes it could display effective countermeasures but the largest point as I have fixed is to display negative outcomes which would be associated with ineffective. I think that I will add effective as well though because that does make it sound better in my opinion.

The 9th point says easy to lean perhaps it should say easy to learn.

Again good catch!!

I wonder if it would be possible to include some names or references to complete the last point,i.e. some of the software available.

I agree I have added a section for software

Last but not least, I would recommend to include more references. It is always advisable for a wiki. By the way, I wonder if ETA is the same as Classifying tree analysis. I have learn about it in the article by Cox and Popken (2008) recommended in this course, and I think it is a very interesting tool in RA. Should it be the case, I would include a mention to this statistical method.

Though you could argue that a classification tree is similar in some ways ETA and Classification tree analysis are different methods of analysis

You can find additional references in this paper and a mention to the rpart algorithm of the stastical software R. Perhaps, you can use it to improve your wiki. Good luck 570acm570acm (talk) 21:08, 9 November 2013 (UTC)

Thank you for the comments!570SJR (talk) 19:47, 13 November 2013 (UTC)[reply]

In the theory section, a probabilistic risk assessment use is stated; however in the methodologies and equation section there is a more deterministic approach described. Distinguishing between instances when a single point probability is used at each intervention point in a deterministic type assessment vs. a stochastic type assessment using probability distributions at each intervention point run with multiple iterations might be beneficial.

I agree that that could be useful but I don’t think that is done in event tree analysis if I am understanding you correctly. I think that you could run simulations to get different probabilities at each event for success or failure but usually the probability of failure comes from a fault tree analysis and then we are able to calculate the probability of success which is 1-pf.

Perhaps examples of the use of varying complexity ETA’s could be added to the page.

I’m not sure what you are saying here, do you mean to add a complex example with many branches or are you talking about using your previously described probability calculation in the ETA?

How is uncertainty addressed when an ETA is conducted?

Generally uncertainty is accepted as this is an estimated approach but in some cases uncertainty is dealt with using fuzzy event tree analysis which I would believe is an entire wiki on its own. What do you guys think ? here is an article on it: http://www.sciencedirect.com/science/article/pii/S0165011498002887#

Huang, D., Chen, T., & Wang, M.-J. J. (2001). A fuzzy set approach for event tree analysis. Fuzzy Sets and Systems, 118(1), 153–165. doi:10.1016/S0165-0114(98)00288-7

Thank you for the feedback!! 570SJR (talk) 20:33, 13 November 2013 (UTC)[reply]