Skip to content

faq 189235210

Billy Charlton edited this page Sep 5, 2018 · 2 revisions

Crucial differences in 'output plans' and 'experienced plans' of the last iteration

by Wolfgang Schieder on 2018-02-27 09:44:56


Dear MATSim community,

in addition to my first, but unfortunately still unanswered question regarding 'Is 'maximumVelocity' ignored in 'equil-mixedTraffic' example?', I am confused about differences in 'output plans' and the 'experienced plans' of the last iteration.

I ran the 'equil-mixedTraffic' with MATSim 0.9.0 (for details see attached ZIP-File containing the input and output directory).

Since no innovation took place and there is only one plan in the agents #7's memory. However, the information in 'output plans' significantly deviates from the information given in 'experienced plans' of the last iteration.

Only in the 'experienced plans' the velocity of 'bicycle' mode seems to be computed correct (15 km/h) leading to plausible travel times!

This is crucial for any evaluation of the 'output plans' file, which seems to be rather useless, if travel times and computed velocity are incorrect!

Please give me response, whether there is a problem within the functionality of MATSim or other reasons for this differences in the output files (or advice how to preferably evaluate the simulation's travel time results - e.g. which file to choose)!

Thanks in advance,

Wolfgang


Comments: 2


Re: Crucial differences in 'output plans' and 'experienced plans' of the last iteration

by Kai Nagel on 2018-02-27 19:16:38

The wording has historically grown, and is quite confusing in this case.

output_ just means that these are files that are generated at the end of the run.

So output_plans is a plans file that was generated at the end of the run. However, it is effectively the input plans to the last iteration. It should be the same as ITERS/it.<finalIteration>/<finalIteration>.plans.xml.gz (except maybe for the score). Input plans may or may not be physically realistic; this is a bit like DNA which may contain junk DNA. For example, activity startTime is always ignored, travelTime is ignored when the mode is executed on the network, etc.

In contrast, experienced_plans is based on what the agents actually did. We should have an output_experiencedPlans, then it would hopefully be a bit clearer.

Quite in general, analysis should be based on output_events . Everything else is for information only.


Re: Crucial differences in 'output plans' and 'experienced plans' of the last iteration

by Wolfgang Schieder on 2018-02-28 08:34:07

Clone this wiki locally