Suppose we have a system designed to deliver services to customers arriving during weekdays. The arrival process is modeled as a Poisson process with an Arrival rate of λ, also we use agent-based modeling with NetLogo to study the behavior of customers. After multiple Observation and replication of the model, the first 8 hours was selected as the warm-up period and the remaining time as the steady-state. If we consider the average length of stay (ALOS) as the crucial output data, how should we handle the initialization bias in this case?
As a workaround, is removing those data sufficient if we take into account the effects of the warm-up period on the ALOS?
Kindly read this work may you can get a strong base about what you are looking for.
https://ieeexplore.ieee.org/document/399839
If you cannot download it, please let me know. I will try to download it from my side.
Regards
Having taken into account the necessary criteria before eliminating bias, as well as testing for statistical consistency within your data, I suppose there's no cause for alarm (it is my opinion).
These works should give you a brief summary of how initialization bias is mitigated in various scenarios with respect to ABMs:
Article Significance of Multi-Site Calibration for Agent-Based Trans...
Article Using Agent-based Simulation and Game Theory to Examine the ...
All the best
Wasim A. Ali and Ebenezer Agbozo thank you both for your helpful answers.
In my case, I was rather reluctant to remove the bias considering the impact of LOS as an essential output; however, after going through verification and validation procedures for the initially developed model, I can confirm that the resulting model is accurate enough to be used for further evaluation.
- Badges
- Science topic
- Similar topics
- Computer Communications (Networks)
- Simulators