Simulation History Database

[abid ghaffar]

Dear Dr. Maarten Sierhuis,

I just need to ask you about Simulation History Database. Whenever we execute Brahms Model, we get Simulation History Database, right. Let's suppose, we do ethnography study for three months for any work practice system and we do simulate scenarios for every single day and we get simulation history database for each day. In this way we may get almost 50 Simulation Run and consequently, we get 50 Simulation History Databases.

When we shall analyse the agents or objects for the same work practice system, how many simulation databases we need to access, the last one or all the previous simulation history databases we created has to be analyse to understand the work practice system or agents or objects.

The point is, I want to analyse the agents behaviour in the work practice system, do I need to access the latest simulation history database or I need all the previous simulation history databases using SQL query. Suggest me, how many simulation history databases do I need to analyse agents or objects behaviour.

Thank you and regards,

Abid Ghaffar

Malaysia

[Maarten Sierhuis]

Dear Abid,

It will all depend on how you change the model for every simulation run. Let’s say you model Day 1 of your observation and run it. Then, the next day you change the model, based on your observation, for ONLY Day 2. So, now your model does NOT include the scenario for Day 1, only Day 2. Etcetera, for all 50 days. Now you indeed have to analyze all 50 databases (which you can do of course with writing a program that loads in and analyzes the databases automatically).

However, if I were you, I would NOT create my models that way. I would change my Day N-1 model, based on my observation of Day N, in such a way that the Day N model includes ALL scenarios from Day 1 to Day N. In this way, I just have to analyze the Day N database, because it included a representation of all possible scenarios from Day 1 to Day N.

Now, the modeling question is of course, how do you create a model that includes a representation of Day 1 till Day N? This is the art of modeling work practice, and does not have one easy answer. My only answer is ABSTRACTION. Find a way to abstract the work practice in such a way that the model represents what people do in “any situation." “Any situation” is in quotes, because it of course depends on how much you have observed.

I hope this helps, although I am afraid I did not answer your question with a straight answer. I could maybe be more helpful if you explain what you are trying to model and what questions you are trying to answer.


Maarten.

> --
> You received this message because you are subscribed to the Google Groups "Brahms Forum" group.
> To unsubscribe from this group and stop receiving emails from it, send an email to brahms-forum...@googlegroups.com.
> To post to this group, send email to brahms...@googlegroups.com.
> Visit this group at http://groups.google.com/group/brahms-forum.
> For more options, visit https://groups.google.com/d/optout.

[abid ghaffar]

Dear Dr. Maarten Sierhuis, 

Thanks for the detailed reply, it answers many questions for my project. Actually, I am trying to figure out human errors in the work practice system. Your detailed reply helps me a lot. I did ethnography study for a university dept. and now detecting human errors in the system.

Thank you and regards,

Abid Ghaffar

[Maarten Sierhuis]

Abid,

It seems that you have been reading my Brahms thesis. In Experiment three, I simulate error handling in the work practice of the Apollo astronauts. This might give you an idea of how to model errors and how to handle them. I'm sure that your medical errors are of a different nature, and I am not sure if the medical professionals notice that they have made an error and then have a practice of how to handle such errors, but still it could give you an idea.

Basically, I modeled the practice of handling of the errors in a separate group. When the astronaut agent would notice an Error event, it would trigger the behavior modeled in this group. The ideas is to classify the errors and the practice of what happens in these cases. By modeling this kind of error handling practice in separate groups, you can have agent inherit this behavior. As your modeling of errors increases, so does the modeling of the error handling behavior by increasing the groups. This helps you in increasing the capability of your model, without having to create different models. 

The trick is of course to classify the different errors and modeling the work practice knowledge of how to handle the different error categories ... a fun exercise in my opinion!

Good luck.

Maarten.