Demands

OD demand distributions for simulation and empirical data.

class micropurc.demands.ODDistribution(origins, dests, weights)

Bases: object

Distribution over origin-destination pairs.

Variables:

• origins (numpy.ndarray) – Origin node indices (internal), shape (M,).

• dests (numpy.ndarray) – Destination node indices (internal), shape (M,).

• weights (numpy.ndarray) – Sampling probabilities, shape (M,), normalized on construction, so a caller may pass raw counts.

Parameters:

• origins (ndarray)

• dests (ndarray)

• weights (ndarray)

origins: ndarray

dests: ndarray

weights: ndarray

property size: int

Number of OD pairs in the distribution.

sample(rng, n_trips)

Draw n_trips OD pairs with replacement, proportional to the weights.

Returns the sampled origins and destinations, each of shape (n_trips,).

Parameters:

• rng (Generator)

• n_trips (int)

Return type:

tuple[ndarray, ndarray]

micropurc.demands.od_from_csv(network, trips_csv, *, origin_col='origin', dest_col='destination', weight_col='sampling_weight', use_node_index=True)

Build an ODDistribution from a CSV file.

Parameters:

• network (Network) – Network for mapping raw node IDs.

• trips_csv (str) – Path to CSV with OD demands.

• origin_col (str) – Column name for origin node IDs.

• dest_col (str) – Column name for destination node IDs.

• weight_col (str | None) – Column holding the weights. When it is None or absent from the file, the OD pairs are weighted equally.

• use_node_index (bool) – Map the raw IDs through network.node_index. With this off, or with no index on the network, the IDs are cast to int and used as internal indices.

Returns:

The origin-destination distribution parsed from the file.

Raises:

ValueError – If the origin or destination column is absent.

Return type:

ODDistribution

micropurc.demands.od_uniform_all_pairs(network, *, allow_self_trips=False)

Uniform distribution over all OD pairs.

Parameters:

• network (Network) – Network whose nodes define the OD universe.

• allow_self_trips (bool) – Keep the pairs whose origin equals their destination.

Returns:

The uniform distribution over the retained OD pairs.

Return type:

ODDistribution