Page 668 - ExtendSim User Guide
P. 668

642 Analysis
Blocks that calculate statistics
Although simulation is a quantified subject, there is still room for analysis and judgmental pro- cedures. This is especially true when your purpose in building a model is to spot trends or iden- tify patterns of behavior, or when you must make an immediate “go, no-go” decision. How statistically precise your models should be depends on several factors, including the nature of the problem, the importance of the decision, the level of risk you are willing to accept, and the sensitivity of the system to the input data.
One of the main benefits of ExtendSim is that it provides several methods for analyzing the results of simulation runs. This chapter covers those techniques, including:
• Blocksforstatisticallyanalyzingdata
• Confidence intervals for predicting the true mean value
• Sensitivity analysis to investigate the impact of making changes
• Optimization to determine the best values for a set of parameters
• Using Stat::Fit to determine the appropriate distribution for a set of data • How to change parameters dynamically
• Plotters for displaying results
• Reporting model details and results
☞ Remember that when you use statistical methods to analyze simulation output, you are per- forming the analysis only on model results, not on the actual system. It is important to ensure that the numbers entered accurately represent the details of the actual system. The significance and relevance of your analysis will depend on how closely the model’s inputs correspond with real-world data (“garbage in = garbage out”).
Blocks that calculate statistics
Many ExtendSim blocks automatically calculate relevant statistics and display them on their Results tab. For example, the Activity block (Item library) reports utilization, average wait, maximum length, and so forth.
You can also use blocks in the Value library, such as the Equation and Math blocks, to perform calculations on model outputs.
As described below, other ExtendSim blocks are specially designed to report statistical infor- mation for particular types of blocks or for items that pass through the block, or to refine statis- tical data collection.
• Activities(Activity,ConveyItem,andTransportblocksintheItemlibrary)
• Mean&Varianceblock(Valuelibrary)
• Queues(Queue,QueueEquation,andQueueMatchingblocksintheItemlibrary) • Resource Item block (Item library)
• Resource Pool block (Item library)
• Mixed blocks (clone drop)
The Statistics block (Value library) accumulates data and calculates statistics using a specified statistical method. Place a Statistics block anywhere in the model and, depending on selections made in its dialog, it will report statistics for the following blocks and types of blocks:
How To

   666   667   668   669   670