Origin-destination travel times

For the computation of departure time choice, one needs information about the trip times as a function of different departure times. [[Different possibilities are discussed in Sec. .]] The implementation that we present here is the arguably simplest method, but it has some caveats for large scale scenarios. [[This will also be discussed in Sec. . ]]

The idea is that one parses the events file, and for each origin, each destination, and each time bin one averages the trip times. This is similar to how the router treats link travel time information. That is, if an individual departs (events flag 6), then this information is stored away somewhere. If the same individual arrives (events flag 0), then the departure time and departure location are retrieved, and the travel time is added by the time bin for the departure time for the corresponding OD pair. Once the complete events file is parsed, the sums are divided to the number of entries as was done for the link travel times. If you assume that you have $T$ time bins, $R$ origins, and $S$ destinations, then this results in $T \times R \times S$ entries.

Task 14.2   Write a script that averages OD travel times into 15-min time bins. Language possibilities are awk or c++/java. As an end result, you should have, for all OD pairs, trip time info for all 15-min time bins. Generate this information for the events file which was obtained by running the traffic microsimulation on

        http://www.matsim.org/files/studies/corridor/teach/0.plans .

Why does the result make sense (or not)?

Note that you have to invent some method to generate OD travel times for time bins for which you have no information.