Validate MATSim Model Besides Comparaing Mode Share from HTS

[jingjunL]

Hi everyone,

I wonder if some additional validation methods exist to validate my simulation with reality?

From the MATSim handbook, I know I should calibrate the model parameters to reach a satisfying mode and distance shares in MATSim compared to our local household travel survey. In addition, I also find some count data in Brussels at some tunnels to validate the traffic volumes. I wonder if there are any other additional approaches that I can use? For instance, I saw some papers using travel time comparisons between Google maps and MATSim. However, I didn't find much practical information online...

Besides, are there any other parameters that may be useful for validation? Such as, I heard some people would use the mode shares of trips within a certain distance (under 1km, 1-2km, more than 5km etc.). Any hint will be helpful!

Thanks in advance, Jingjun

[sebhoerl]

See for instance the appendix of this paper: Hörl, S., Becker, F., Axhausen, K.W., 2021. Simulation of price, customer behaviour and system impact for a cost-covering automated taxi system in Zurich. Transportation Research Part C: Emerging Technologies 123, 102974.

We were calibrating

• Mode share by distance
• Mean / Median / Q10 / Q90 car travel time by distance
• Distribution of car travel times

Here are some additional ideas: Exploring accelerated evolutionary parameter search for iterative large-scale transport simulations in a new calibration testbed

For New York they used SPSA to calibrate capacities and compared with count data: He, B.Y., Zhou, J., Ma, Z., Wang, D., Sha, D., Lee, M., Chow, J.Y.J., Ozbay, K., 2021. A validated multi-agent simulation test bed to evaluate congestion pricing policies on population segments by time-of-day in New York City. Transport Policy 101, 145–161.

[jingjunL]

See for instance the appendix of this paper: Hörl, S., Becker, F., Axhausen, K.W., 2021. Simulation of price, customer behaviour and system impact for a cost-covering automated taxi system in Zurich. Transportation Research Part C: Emerging Technologies 123, 102974.

We were calibrating

• Mode share by distance
• Mean / Median / Q10 / Q90 car travel time by distance
• Distribution of car travel times

Here are some additional ideas: Exploring accelerated evolutionary parameter search for iterative large-scale transport simulations in a new calibration testbed

For New York they used SPSA to calibrate capacities and compared with count data: He, B.Y., Zhou, J., Ma, Z., Wang, D., Sha, D., Lee, M., Chow, J.Y.J., Ozbay, K., 2021. A validated multi-agent simulation test bed to evaluate congestion pricing policies on population segments by time-of-day in New York City. Transport Policy 101, 145–161.

Hi Sebastian,

Thanks for the suggestions and they are pretty helpful :)

Everyone has his way of calibrating the model. I will not close this issue in case someone else has further suggestions!