Dealing with Replications Time Series Considerations

9 Dealing with Replications Time Series Considerations 

One of the major features of any simulation study is its abihty to rephcate results easily and quickly. With a sufficient number of rephcations, we can approximate the underlying system parameters (e g., means, variances). Rephcation is a critical 

When replicating results for a given combination of parameters and experimental policies, we must realize that we are faced by an M x N matrix (M replications by N time periods). Rather than carrying out ARIMA and Outlier Detection for each replication, an alternative approach is to generate a new 1 x N matrix, where each cell consists of the mean value for the values from the M replications. This Mean Time Series (MTS) becomes the focal point for application of time series analysis. Further, with sufficient rephcations, we can assume that the values contained within this MTS are unbiased estimators of the true system values for that time period. 

---