benchmark-urbanism / cityseer-api Goto Github PK

Display the geometries for segments when plotting.

Requires a bit more thought about whether to further formalise and refine the plotting options...

The crawl procedure has been removed from the graph consolidation method and replaced with a by_neighbours strategy. This strategy better preserves topology.

A subsequent improvement would be to split parallel geometries where nodes are out of lock-step. This may remove the need for decomposition steps prior to final consolidation, which might result in cleaner and quicker consolidation.

Progress bars are currently difficult with parallel execution in numba.

See krishnanlab/PecanPy#5 / numba/numba#4267

Forthcoming enhancements to numba may add configurable chunk-sizes for prange which may therefore allow for a more polished parallel mode progress bar, ideally with callback to objmode for invocation of something like tqdm.

See numba/numba#6025

For now, non-sequential progress bar updates and CPU watching will have to do!

Zero-length node maps are caught and reported by the checks module, though it may be good to flag this at an earlier stage of graph creation.

Describe the bug
When using a network that has coordinate values below 0, like in case of EPSG 5514, node_centrality raises an error as you have a check there for this case.

To Reproduce
Steps to reproduce the behavior:

import geopandas
import momepy
import networkx as nx

from cityseer.tools import graphs
from cityseer.metrics import networks

df = geopandas.read_file(momepy.datasets.get_path("bubenec"), layer="streets").to_crs(5514)
# get compatible graph
G = momepy.gdf_to_nx(df)
nx.set_node_attributes(G, {n: {"x":n[0], "y":n[1]} for n in G.nodes})
nx.set_edge_attributes(G, {(i, v, k): {"geom":a["geometry"]} for i, v, k, a in G.edges(data=True, keys=True)})

nodes_gdf, network_structure = graphs.network_structure_from_nx(G, crs=df.crs)

nodes_gdf = networks.node_centrality(
    measures=["node_harmonic", "node_betweenness"],
    network_structure=network_structure,
    nodes_gdf=nodes_gdf,
    distances=[400],
)

---------------------------------------------------------------------------
ValueError                                Traceback (most recent call last)
/Users/martin/Git/momepy/Untitled-1.ipynb Cell 10 in <cell line: 1>()
----> 1 nodes_gdf = networks.node_centrality(
      2     measures=["node_harmonic", "node_betweenness"],
      3     network_structure=network_structure,
      4     nodes_gdf=nodes_gdf,
      5     distances=[400],
      6 )

File ~/mambaforge/envs/geo_dev/lib/python3.10/site-packages/cityseer/metrics/networks.py:499, in node_centrality(measures, network_structure, nodes_gdf, distances, betas, jitter_scale, angular, min_threshold_wt)
    423 def node_centrality(
    424     measures: Union[list[str], tuple[str]],
    425     network_structure: structures.NetworkStructure,
   (...)
    431     min_threshold_wt: float = MIN_THRESH_WT,
    432 ) -> gpd.GeoDataFrame:
    433     r"""
    434     Compute node-based network centrality.
    435 
   (...)
    497 
    498     """
--> 499     network_structure.validate()
    500     _distances, _betas = pair_distances_betas(distances, betas, min_threshold_wt)
    501     # see centrality.local_centrality for integrity checks on closeness and betweenness keys
    502     # typos are caught below

File ~/mambaforge/envs/geo_dev/lib/python3.10/site-packages/numba/experimental/jitclass/boxing.py:61, in _generate_method.<locals>.wrapper(*args, **kwargs)
     59 @wraps(func)
     60 def wrapper(*args, **kwargs):
---> 61     return method(*args, **kwargs)

File ~/mambaforge/envs/geo_dev/lib/python3.10/site-packages/cityseer/structures.py:92, in validate()
     90     raise ValueError("X, Y and 'live' arrays are not the same length")
     91 if not np.all(np.isfinite(self.xs)) or not np.all(self.xs >= 0):
---> 92     raise ValueError("Missing or invalid start x data encountered.")
     93 if not np.all(np.isfinite(self.ys)) or not np.all(self.ys >= 0):
     94     raise ValueError("Missing or invalid start y data encountered.")

ValueError: Missing or invalid start x data encountered.

Expected behavior
The computation should normally work, this will not be the only use case where negative projected coordinates appear.

Hi,

When I try to import io using

from cityseer.tools import graphs, io

I am getting

TypeError: 'numpy._DTypeMeta' object is not subscriptable


Output from spyder call 'get_namespace_view':
/Users/mohammadalattar/opt/anaconda3/lib/python3.9/site-packages/spyder_kernels/utils/nsview.py:443: UserWarning: legacy printing option can currently only be '1.13', '1.21', or `False`
  np.set_printoptions(**np_printoptions

Hello,

I've successfully installed the cityseer, but when I tried to import graphs, io using

from cityseer.tools import graphs, io

an error occurred.
How could I handle it?
Many thanks

Betweenness assigns a node to the (first) path on the shortest route.

Therefore, the node will be assigned to the first out of multiple equidistant routes from A to B.

When working with shortest or simplest paths, it is highly unlikely for exactly equal paths to exist; nevertheless, it would be nice to figure out a method for dispersing node assignments for routes that are approximately equal to each other. e.g. Manhattan-like grids where numerous approximately equal routes exist from A to B.

A drawback to the above is that this may substantially complicate the underlying shortest path algorithm because multiple predecessors have to be tracked if they are within some or another tolerance of each other...

A tempting strategy may be to modify the usual shortest path algorithm where updates to the routes in the shortest-path tree are randomly executed if within a small tolerance of the existing shortest path to the same node in the tree. e.g. assuming a 2% tolerance, then a newly found route of 98m would randomly override / not override an existing shortest-route of 100m. The problem with this approach is that all subsequent calculations for shortest / between paths will follow this same route... i.e. it is the same problem as before, just that the route isn't necessarily the first option.

Consider adding a heatmap plot function.

This is useful for large graphs.

It is worth comparing performance if switching from a simple pythagorean spatial filter to the shapely rtree filter.

Hi,

I've noticed that cityseer pins requires-python = ">=3.10, <3.12". What is blocking support of Python 3.12?

Edit: The reason I am asking is that pip install cityseer tries to pull 1.2.1 because that seems to be the last one without that pin. And that version is broken due to #39. Even if it wasn't, it is quite out of date :).

Return column with zeros instead of nothing if landuse / stats label does not exist.

Remove unnecessary smaller distance thresholds for distance to landuse measures

Describe the bug
when running the shortest path node centrality the progress does not update. Running on VS code on windows 10

To Reproduce
Steps to reproduce the behavior:

1. Go to '...'
2. Click on '....'
3. Scroll down to '....'
4. See error

Expected behavior
A clear and concise description of what you expected to happen.

Additional context
Add any other context about the problem here.

@martinfleis

I'm wondering about adding a conversion function for a momepy street network GeoDataFrame to cityseer compatible networkx MultiGraph .

Here is an initial sketch of what this might look like: momepy_to_cityseer

This would mainly be intended for people who want to:

• run street network centralities on larger street networks
• calculate network centralities using cityseer's methods which adapt the shortest-path algorithms to work for some situations which aren't necessarily covered by e.g. networkX (e.g. shortest path sidestepping issue)
• calculate landuse accessibilities over the street network (weighted and unweighted)
• calculate mixed-uses over the street network (weighted and unweighted)
• aggregate statistics (e.g. building form metrics from momepy) using the street network as a backbone

I'm curious what your thoughts are on things to take into account so that this is useful, e.g.

• would there be use cases where people might want to do a round-trip - e.g. momepy GeoDataFrame to cityseer then back to momepy? If so, the edge ids would need to be chained through. The drawback to a round-trip is that network cleaning or decomposition can't be done without losing the association.
• the current conversion function doesn't take primal or dual into account, it just builds a network from the edges. This is probably a non-issue unless doing a round-trip as people can simply manipulate the graphs either end depending on what they want to do.
• would there be more interest in going straight from a GeoDataFrame or would there be cases where it is better to go from a momepy formatted networkx Graph?

There are probably other things but that's all that comes to mind for now...

Is your feature request related to a problem? Please describe.

Add type checking for statistical data aggregation - e.g. catch None and other invalid (per numpy / numba) data types

Numba is introducing support for functions as first-class types. This means that the workflows in the core centrality and landuse methods can be parcelled into functions which can then be placed in a list depending on selected options.