Analyzing Systems with Many Dependent Components Through the Complex Systems Point of View.

author.DisplayName
Industrial Engineering, Ariel University, ישראל

Complex Systems, and Network Sciences are two adjacent fields which emerged from statistical physics. These fields study the behavior of systems with a large set of influences between the many sub-entities of a larger system.

The traditional Industrial Engineering methods assume in many cases, that much of the measures in an organization are Normally distributed. For example, sales, demands, prices of materials or customer`s preferences, are all assumed to follow a Normal distribution. But it is often forgotten, that one of the preconditions for a process that generates a Normally distributed measure, is that this measure is a sum of many independent effects from the sub components of the system. In our interconnected world, this assumption is frequently violated. Customer`s demands are driven by their communication paths. Bad experience of a product might be shared over the internet. Material`s prices, driven by their demand and supply, change as information on a supplier`s difficulties, spreads and shifts the prices of other firms. Thus, the assumption of independence might sometimes be highly incorrect. Here the field of Network Science comes in hand.

The Goal of the workshop is to first awareness. This will be achieved by demonstrating the need for a careful use of a Normal distribution, through visual simulative animations. These will demonstrate the results when the independence assumption is violated by a positive feedback loop, as opposed to a case of independence.

Then, the workshop will continue into more applied directions. This will be done by showing usable tools and methodologies that can be used as a first step in the analysis of a correlated system. These tools will include Agent Based Simulations, and Network Science concepts and their use through case studies.

We will present the concept of "emergence", in complex systems, which are almost impossible evets in a world regulated by Normally distributed processes, but which are likely to occur when positive feedback exists.

Description: The workshop will demonstrate through case studies, where Network Science tools can be used to study interconnected systems.

It will introduce a simple set of methodologies, to analyze a non-independent system, and is planned to enable some "take home" experience and message.

The workshop is planned for an audience of Industrial Engineers, and is presented as an introductory level. Mainly, it intends to provide a basic, clear and practical understanding of the limitations and cautions that need to be taken when using common tools such as ANOVA tables, Normality based predictions, and similar statistical tools, in a non-independent system.





החברה המארגנת: ארטרא בע"מ, רחוב יגאל אלון 94 תל אביב 6109202 טלפון: 03-6384444, פקס: 6384455–03
iem@ortra.com מייל לשאלות





Powered by Eventact EMS