Simulation-Run-Statistics
- 1. Simulation Run Statistics Sachin dheke 013-333 Sadiksha kafle 013-334 Sagar khadka 013-335
- 2. INTRODUCTION Method to handle problems that arise in measuring statistics from simulation runs Method used to analyze simulation results. Two assumption made to establish the confidence levels i) Observation are independent. ii) Distribution is stationary. Many statistics do not meet these condition.
- 3. Example • Single server system(denoted by MM1) M= inter-arrival time is distribution exponentially M= the service time is distributed exponentially 1- one server. • First-in , first-out with no priority. Objective: • To measure the mean waiting time.
- 4. Mean Waiting Time • Simplest approach to estimate mean waiting time. • Usual formula to estimate mean value: - Sample Mean - individual waiting times • Calculated waiting time is dependent • Data are autocorrelated • Varience of autocorrelated data is not related to population variance • Positive term is added for autocorrelation but may be negetive for other system. n i n i xnx 1 1 )( )(nx i x
- 5. Another Problem • Distribution not stationary. • Early arrival obtain service quickly so sample mean including it will be biased • Biased die out as simulation length extends and sample size increases.
- 7. Description • Figure based on theoretical results • Show dependency between expected value of sample mean and sample length for M/M/1 system • Server utilization=0.9 • Steady state mean=8.1 • Mean value biased below steady state mean • As sample size increase bias diminishes but even sample =2000 mean only reached 95% of steady state value.
- 8. THANK YOU