Compute travel times with a detailed breakdown of the routing results#

Detailed itineraries#

In case you are interested in more detailed routing results, you can use DetailedItinerariesComputer. In contrast to TravelTimeMatrixComputer, it reports individual trip segments, and possibly multiple alternative routes for each trip.

As such, DetailedItinerariesComputer’s output is structured in a different way, too. It provides one row per trip segment, multiple trip segments together constitute a trip option, of which there might be several per from_id/to_id pair. The results also include information on the public transport routes (e.g., bus line numbers) used on the trip, as well as a shapely.geometry for each segment.

Detailed itineraries are computationally expensive

Computing detailed itineraries is significantly more time-consuming than calculating simple travel times. As such, think twice whether you actually need the detailed information output from this function, and how you might be able to limit the number of origins and destinations you need to compute.

For the examples below, to reduce computation effort, we use a sample of 3 origin points and one single destination (the railway station) in our sample data of Helsinki.

import geopandas
import r5py
import shapely

population_grid = geopandas.read_file(
RAILWAY_STATION = shapely.Point(24.941521, 60.170666)

transport_network = r5py.TransportNetwork(,
import datetime
import r5py

origins = population_grid.sample(3).copy()
origins.geometry = origins.geometry.centroid

destinations = geopandas.GeoDataFrame(
            "id": [1],
            "geometry": [RAILWAY_STATION]

detailed_itineraries_computer = r5py.DetailedItinerariesComputer(
    transport_modes=[r5py.TransportMode.TRANSIT, r5py.TransportMode.WALK],

Snap to network

If you read the code block above especially carefully, you may have noticed that we added an option snap_to_network=True to DetailedItinerariesComputer. This option does exactly what it says on the outside: it attempts to snap all origin and destination points to the transport network before routing. This can help with points that come to lie in an otherwise inaccessible area, such as a fenced area, a swamp, or the middle of a lake.

For a detailed description of the functionality, see the Advanced use page.

travel_details = detailed_itineraries_computer.compute_travel_details()
from_id to_id option segment transport_mode departure_time distance travel_time wait_time route geometry
0 8 1 0 0 TransportMode.WALK NaT 2462.124000 0 days 00:42:06 NaT None LINESTRING (24.91213 60.15755, 24.91224 60.157...
1 8 1 1 0 TransportMode.WALK 2022-02-22 08:38:53 141.745000 0 days 00:02:30 0 days 00:00:00 None LINESTRING (24.91213 60.15755, 24.91224 60.157...
2 8 1 1 1 TransportMode.TRAM 2022-02-22 08:43:00 3395.346410 0 days 00:08:00 0 days 00:01:46 9 LINESTRING (24.91168 60.15681, 24.91187 60.156...
3 8 1 1 2 TransportMode.WALK 2022-02-22 08:52:00 702.036000 0 days 00:11:59 0 days 00:00:00 None LINESTRING (24.94150 60.17067, 24.94159 60.170...
4 8 1 2 0 TransportMode.WALK 2022-02-22 08:35:37 340.264000 0 days 00:05:49 0 days 00:00:00 None LINESTRING (24.91213 60.15755, 24.91224 60.157...
... ... ... ... ... ... ... ... ... ... ... ...
540 23 1 59 1 TransportMode.TRAM 2022-02-22 08:48:00 734.486981 0 days 00:01:00 0 days 00:04:38 9 LINESTRING (24.92300 60.16280, 24.92274 60.163...
541 23 1 59 2 TransportMode.WALK 2022-02-22 08:50:00 1099.593000 0 days 00:18:49 0 days 00:00:00 None LINESTRING (24.94150 60.17067, 24.94159 60.170...
542 23 1 60 0 TransportMode.WALK 2022-02-22 08:32:12 263.831000 0 days 00:04:29 0 days 00:00:00 None LINESTRING (24.92127 60.16452, 24.92128 60.164...
543 23 1 60 1 TransportMode.TRAM 2022-02-22 08:38:00 2758.085641 0 days 00:06:00 0 days 00:01:19 9 LINESTRING (24.92300 60.16280, 24.92274 60.163...
544 23 1 60 2 TransportMode.WALK 2022-02-22 08:45:00 51.910000 0 days 00:00:55 0 days 00:00:00 None LINESTRING (24.94150 60.17067, 24.94159 60.170...

545 rows × 11 columns

As you can see, the result contains much more information than earlier. Depending on your screen size, you might even have to scroll further right to see all columns.

Especially in the case of public transport routes, or when choosing a list of different transport_modes, also the table structure of the results is more complex: For each origin/destination pair, one or more possible option is reported, which in turn can consist of one or more segments. Both options and segments are numbered sequentially, starting at 0.

Each segment, then, represents one row in the results table, and provides information about the transport mode used for a segment, time travelled, possible wait time (before the departure of a public transport vehicle), the route (e.g., bus number, metro line), and finally a line geometry representing the travelled path.

See the following table for a complete list of columns returned by DetailedItinerariesComputer.compute_travel_details():

from_id (same type as origins["id"])

the origin of the trip this segment belongs to

to_id (same type as destinations["id"])

the destination of the trip this segment belongs to

option (int)

sequential number enumerating the the different trip options found. Each trip option consists of one or more trip segments. (starts with 0)

segment (int)

sequential number enumerating the segments the current trip option consists of. (starts with 0)

transport_mode (r5py.TransportMode)

the transport mode used on the current segment

departure_time (datetime.datetime)

the departure date and time of the public transport vehicle used for the current segment; NaT in case of other modes of transport

distance (float)

the distance travelled on the current segment, in metres. For public transport, see note below.

travel_time (datetime.timedelta)

The time spent travelling on the current segment

wait_time (datetime.timedelta)

if the current segment is a public transport vehicle: wait time between the arrival of the previous trip segment and the departure of the current segment.

route (str)

if the current segment is a public transport vehicle: the route number (or other id), as specified in the input GTFS data set, e.g. bus numbers, metro line names

geometry (shapely.LineString)

the path travelled on the current segment. For public transport, see note below.

Geometries of public transport routes, and distances travelled

The default upstream version of R⁵ is compiled with com.conveyal.r5.transit.TransitLayer.SAVE_SHAPES = false, which improves performance by not reading the geometries included in GTFS data sets.

As a consequence, the geometry reported by DetailedItinerariesComputer are straight lines in-between the stops of a public transport line, and do not reflect the actual path travelled in public transport modes.

With this in mind, r5py does not attempt to compute the distance of public transport segments if SAVE_SHAPES = false, as distances would be very crude approximations, only. Instead it reports NaN/None.

The Digital Geography Lab maintains a patched version of R⁵ in its GitHub repositories. If you want to refrain from compiling your own R⁵ jar, but still would like to use detailed geometries of public transport routes, follow the instructions in Advanced use.

Visualise travel details#

It’s not difficult to plot the detailed routes in a map, however, a couple more steps are needed than with simple travel times. GeoDataFrame.explore() cannot handle the column types r5py.TransportMode and datetime.timedelta - the conversion is quick and easy, though:

travel_details["mode"] = travel_details.transport_mode.astype(str)
travel_details["travel time (min)"] = travel_details.travel_time.apply(
    lambda t: round(t.total_seconds() / 60.0, 2)
travel_details["trip"] = travel_details.apply(
    lambda row: f"{row.from_id} → railway station",

detailed_routes_map = (
            "travel time (min)",
        tooltip=["trip", "option", "segment", "mode", "travel time (min)", "distance"],

Let’s also add the origins and the destination to the map:

import folium
import folium.plugins
import pandas


points = geopandas.GeoDataFrame(
        {"id": detailed_itineraries_computer.od_pairs["id_origin"].unique()}
points.geometry = points.geometry.to_crs("EPSG:3875").centroid.to_crs("EPSG:4326")

    lambda row: (
            (row["geometry"].y, row["geometry"].x),

Make this Notebook Trusted to load map: File -> Trust Notebook

Export the detailed routes#

If you want to further analyse the resulting routes, for instance, in a desktop GIS, you can export the GeoDataFrame to a wide range of file formats, using the to_file() method.

Note that many geospatial file formats do not support datetime.timedelta columns, or columns with custom objects, such as the r5py.TransportMode data. Similar to the above example, with a few simple steps we can convert the values accordingly:

travel_details["transport_mode"] = travel_details.transport_mode.astype(str)
travel_details["travel time (min)"] = travel_details.travel_time.apply(
    lambda t: round(t.total_seconds() / 60.0, 2)
travel_details["wait time (min)"] = travel_details.wait_time.apply(
    lambda t: round(t.total_seconds() / 60.0, 2)

# keep all columns except travel time and wait time (which we renamed to
# reflect the unit of measurement)
travel_details = travel_details[
        "travel time (min)",
        "wait time (min)",