Highway2Vec Embedder¶

In [1]:

Copied!

from IPython.display import display

from srai.plotting import plot_numeric_data, plot_regions
from IPython.display import display

from srai.plotting import plot_numeric_data, plot_regions

Get an area to embed¶

In [2]:

Copied!

from srai.regionalizers import geocode_to_region_gdf

area_gdf = geocode_to_region_gdf("Wrocław, PL")
plot_regions(area_gdf, tiles_style="CartoDB positron")
from srai.regionalizers import geocode_to_region_gdf

area_gdf = geocode_to_region_gdf("Wrocław, PL")
plot_regions(area_gdf, tiles_style="CartoDB positron")

Out[2]:

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

Regionalize the area with a regionalizer¶

In [3]:

Copied!





from srai.regionalizers import H3Regionalizer

regionalizer = H3Regionalizer(9)
regions_gdf = regionalizer.transform(area_gdf)
print(len(regions_gdf))
display(regions_gdf.head(3))
plot_regions(regions_gdf, tiles_style="CartoDB positron")
from srai.regionalizers import H3Regionalizer

regionalizer = H3Regionalizer(9)
regions_gdf = regionalizer.transform(area_gdf)
print(len(regions_gdf))
display(regions_gdf.head(3))
plot_regions(regions_gdf, tiles_style="CartoDB positron")

	geometry
region_id
891e20421cbffff	POLYGON ((16.94046 51.16970, 16.94050 51.16802...
891e2042a2fffff	POLYGON ((17.00658 51.13133, 17.00661 51.12965...
891e204e557ffff	POLYGON ((17.05566 51.06461, 17.05570 51.06293...

Out[3]:

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

Download a road infrastructure for the area¶

In [4]:

Copied!

from srai.loaders import OSMNetworkType, OSMWayLoader

loader = OSMWayLoader(OSMNetworkType.DRIVE)
nodes_gdf, edges_gdf = loader.load(area_gdf)

display(nodes_gdf.head(3))
display(edges_gdf.head(3))

ax = edges_gdf.plot(linewidth=1, figsize=(12, 7))
nodes_gdf.plot(ax=ax, markersize=3, color="red")
from srai.loaders import OSMNetworkType, OSMWayLoader

loader = OSMWayLoader(OSMNetworkType.DRIVE)
nodes_gdf, edges_gdf = loader.load(area_gdf)

display(nodes_gdf.head(3))
display(edges_gdf.head(3))

ax = edges_gdf.plot(linewidth=1, figsize=(12, 7))
nodes_gdf.plot(ax=ax, markersize=3, color="red")

/opt/hostedtoolcache/Python/3.10.15/x64/lib/python3.10/site-packages/srai/loaders/osm_way_loader/osm_way_loader.py:229: FutureWarning: The clean_periphery argument has been deprecated and will be removed in the v2.0.0 release. Future behavior will be as though clean_periphery=True. See the OSMnx v2 migration guide: https://github.com/gboeing/osmnx/issues/1123
  G_directed = ox.graph_from_polygon(
/opt/hostedtoolcache/Python/3.10.15/x64/lib/python3.10/site-packages/osmnx/_overpass.py:359: FutureWarning: `settings.timeout` is deprecated and will be removed in the v2.0.0 release: use `settings.requests_timeout` instead. See the OSMnx v2 migration guide: https://github.com/gboeing/osmnx/issues/1123
  overpass_settings = _make_overpass_settings()

/opt/hostedtoolcache/Python/3.10.15/x64/lib/python3.10/site-packages/osmnx/_overpass.py:369: FutureWarning: `settings.timeout` is deprecated and will be removed in the v2.0.0 release: use `settings.requests_timeout` instead. See the OSMnx v2 migration guide: https://github.com/gboeing/osmnx/issues/1123
  yield _overpass_request(data={"data": query_str})
/opt/hostedtoolcache/Python/3.10.15/x64/lib/python3.10/site-packages/osmnx/_overpass.py:451: FutureWarning: `settings.timeout` is deprecated and will be removed in the v2.0.0 release: use `settings.requests_timeout` instead. See the OSMnx v2 migration guide: https://github.com/gboeing/osmnx/issues/1123
  this_pause = _get_overpass_pause(overpass_endpoint)

/opt/hostedtoolcache/Python/3.10.15/x64/lib/python3.10/site-packages/srai/loaders/osm_way_loader/osm_way_loader.py:237: FutureWarning: The `get_undirected` function is deprecated and will be removed in the v2.0.0 release. Replace it with `convert.to_undirected` instead. See the OSMnx v2 migration guide: https://github.com/gboeing/osmnx/issues/1123
  G_undirected = ox.utils_graph.get_undirected(G_directed)

	y	x	street_count	highway	ref	geometry
osmid
95584835	51.083111	17.049513	4	NaN	NaN	POINT (17.04951 51.08311)
95584841	51.084699	17.064367	3	NaN	NaN	POINT (17.06437 51.08470)
95584850	51.083328	17.035057	4	NaN	NaN	POINT (17.03506 51.08333)

	oneway	lanes-1	lanes-2	lanes-3	lanes-4	lanes-5	lanes-6	lanes-7	lanes-8	lanes-9	...	bicycle-official	lit-yes	lit-no	lit-sunset-sunrise	lit-24/7	lit-automatic	lit-disused	lit-limited	lit-interval	geometry
feature_id
0	0	0	1	0	0	0	0	0	0	0	...	0	1	0	0	0	0	0	0	0	LINESTRING (17.04951 51.08311, 17.04947 51.083...
1	0	0	1	0	0	0	0	0	0	0	...	0	1	0	0	0	0	0	0	0	LINESTRING (17.04951 51.08311, 17.04933 51.083...
2	0	0	1	1	1	0	0	0	0	0	...	0	1	0	0	0	0	0	0	0	LINESTRING (17.05357 51.08301, 17.05335 51.082...

3 rows × 219 columns

Out[4]:

<Axes: >

No description has been provided for this image

Find out which edges correspond to which regions¶

In [5]:

Copied!

from srai.joiners import IntersectionJoiner

joiner = IntersectionJoiner()
joint_gdf = joiner.transform(regions_gdf, edges_gdf)
joint_gdf
from srai.joiners import IntersectionJoiner

joiner = IntersectionJoiner()
joint_gdf = joiner.transform(regions_gdf, edges_gdf)
joint_gdf

Out[5]:


region_id	feature_id
891e204e557ffff	4190
	4202
	4201
	4199
891e2042697ffff	8901
...	...
891e20406b3ffff	8286
	3362
	6578
891e2055a43ffff	7511
891e2055a43ffff	8970

15722 rows × 0 columns

Get the embeddings for regions based on the road infrastructure¶

In [6]:

Copied!

from pytorch_lightning import seed_everything

from srai.embedders import Highway2VecEmbedder

seed_everything(42)

embedder = Highway2VecEmbedder()
embeddings = embedder.fit_transform(regions_gdf, edges_gdf, joint_gdf)
embeddings
from pytorch_lightning import seed_everything

from srai.embedders import Highway2VecEmbedder

seed_everything(42)

embedder = Highway2VecEmbedder()
embeddings = embedder.fit_transform(regions_gdf, edges_gdf, joint_gdf)
embeddings

Seed set to 42

GPU available: False, used: False

TPU available: False, using: 0 TPU cores

IPU available: False, using: 0 IPUs

HPU available: False, using: 0 HPUs

/opt/hostedtoolcache/Python/3.10.15/x64/lib/python3.10/site-packages/pytorch_lightning/trainer/connectors/logger_connector/logger_connector.py:75: Starting from v1.9.0, `tensorboardX` has been removed as a dependency of the `pytorch_lightning` package, due to potential conflicts with other packages in the ML ecosystem. For this reason, `logger=True` will use `CSVLogger` as the default logger, unless the `tensorboard` or `tensorboardX` packages are found. Please `pip install lightning[extra]` or one of them to enable TensorBoard support by default

  | Name    | Type       | Params
---------------------------------------
0 | encoder | Sequential | 16.0 K
1 | decoder | Sequential | 16.2 K
---------------------------------------
32.1 K    Trainable params
0         Non-trainable params
32.1 K    Total params
0.128     Total estimated model params size (MB)

/opt/hostedtoolcache/Python/3.10.15/x64/lib/python3.10/site-packages/pytorch_lightning/trainer/connectors/data_connector.py:441: The 'train_dataloader' does not have many workers which may be a bottleneck. Consider increasing the value of the `num_workers` argument` to `num_workers=3` in the `DataLoader` to improve performance.

`Trainer.fit` stopped: `max_epochs=10` reached.

/opt/hostedtoolcache/Python/3.10.15/x64/lib/python3.10/site-packages/srai/embedders/highway2vec/embedder.py:75: FutureWarning: The behavior of array concatenation with empty entries is deprecated. In a future version, this will no longer exclude empty items when determining the result dtype. To retain the old behavior, exclude the empty entries before the concat operation.
  embeddings_joint = joint_gdf.join(embeddings_df)

Out[6]:

	0	1	2	3	4	5	6	7	8	9	...	20	21	22	23	24	25	26	27	28	29
region_id
891e2040007ffff	-0.500876	-0.170577	0.175679	-0.138017	0.351574	-0.137988	0.097659	-0.601147	-0.241708	0.738275	...	0.326857	-0.527040	0.132806	0.072183	-0.015617	0.040504	0.171830	0.370400	0.146430	0.348382
891e2040013ffff	-0.096451	-0.105448	0.285260	-0.296229	0.622387	-0.110835	0.036012	-0.672212	-0.479695	0.815160	...	0.413562	-0.754025	0.198783	-0.328841	-0.017702	0.092769	0.455044	0.404475	0.166973	0.372385
891e2040017ffff	-0.284640	-0.256166	0.183288	-0.233499	0.446882	-0.211849	0.230594	-0.792736	-0.233340	0.618573	...	0.195669	-0.576084	0.197247	0.038464	0.048569	0.083445	0.470839	0.281041	0.138333	0.396957
891e2040023ffff	-0.241505	-0.363091	0.301251	-0.026235	0.519482	-0.183924	0.170449	-0.417670	-0.102240	0.612353	...	0.133085	-0.501896	0.179845	0.035775	-0.074151	0.209274	0.391761	0.097197	-0.033675	0.442482
891e2040027ffff	-0.561028	-0.269774	0.069739	-0.329874	0.631586	0.011030	0.127321	-0.714779	-0.320608	0.769084	...	-0.003671	-0.611373	0.175998	-0.074924	-0.033651	0.277068	0.207851	0.231221	0.038559	0.586329
...	...	...	...	...	...	...	...	...	...	...	...	...	...	...	...	...	...	...	...	...	...
891e2055bc3ffff	-0.499624	-0.015107	0.130929	-0.192005	0.520919	-0.350832	0.242332	-0.776085	-0.352089	0.773911	...	0.466479	-0.584363	-0.091274	-0.211894	-0.304317	0.270350	0.190800	0.388361	0.321336	0.761356
891e2055bc7ffff	-0.353990	0.102986	0.096897	-0.284011	0.792185	-0.537846	0.339987	-0.508364	-0.397602	0.779856	...	0.277303	-0.611381	-0.034055	-0.250024	-0.416371	0.289269	0.047325	0.326037	0.210747	0.463528
891e2055bcbffff	-0.649712	-0.050774	0.189476	-0.130992	0.493045	-0.208345	0.289957	-1.025107	-0.309696	0.747634	...	0.634918	-0.514004	-0.267735	-0.272899	-0.313647	0.365553	0.275041	0.450990	0.498539	1.033792
891e205a967ffff	-0.605251	0.114015	0.175078	-0.281711	0.336612	0.154542	-0.026165	-0.365951	-0.396474	0.531186	...	-0.299310	-0.325677	0.356714	0.028953	0.215758	-0.004623	0.233533	0.279212	0.088734	0.554686
891e205a9a7ffff	-1.404263	-1.382761	0.503229	-1.089563	0.939539	-0.933452	0.222206	-0.786773	-0.500759	1.735679	...	0.565642	-0.589892	-0.091554	0.035516	-0.432250	0.239600	0.643934	1.037692	0.163673	0.953123

2034 rows × 30 columns

In [7]:

Copied!

from sklearn.cluster import KMeans

clusterizer = KMeans(n_clusters=5, random_state=42)
clusterizer.fit(embeddings)

embeddings["cluster"] = clusterizer.labels_
embeddings
from sklearn.cluster import KMeans

clusterizer = KMeans(n_clusters=5, random_state=42)
clusterizer.fit(embeddings)

embeddings["cluster"] = clusterizer.labels_
embeddings

Out[7]:

	0	1	2	3	4	5	6	7	8	9	...	21	22	23	24	25	26	27	28	29	cluster
region_id
891e2040007ffff	-0.500876	-0.170577	0.175679	-0.138017	0.351574	-0.137988	0.097659	-0.601147	-0.241708	0.738275	...	-0.527040	0.132806	0.072183	-0.015617	0.040504	0.171830	0.370400	0.146430	0.348382	1
891e2040013ffff	-0.096451	-0.105448	0.285260	-0.296229	0.622387	-0.110835	0.036012	-0.672212	-0.479695	0.815160	...	-0.754025	0.198783	-0.328841	-0.017702	0.092769	0.455044	0.404475	0.166973	0.372385	3
891e2040017ffff	-0.284640	-0.256166	0.183288	-0.233499	0.446882	-0.211849	0.230594	-0.792736	-0.233340	0.618573	...	-0.576084	0.197247	0.038464	0.048569	0.083445	0.470839	0.281041	0.138333	0.396957	3
891e2040023ffff	-0.241505	-0.363091	0.301251	-0.026235	0.519482	-0.183924	0.170449	-0.417670	-0.102240	0.612353	...	-0.501896	0.179845	0.035775	-0.074151	0.209274	0.391761	0.097197	-0.033675	0.442482	4
891e2040027ffff	-0.561028	-0.269774	0.069739	-0.329874	0.631586	0.011030	0.127321	-0.714779	-0.320608	0.769084	...	-0.611373	0.175998	-0.074924	-0.033651	0.277068	0.207851	0.231221	0.038559	0.586329	1
...	...	...	...	...	...	...	...	...	...	...	...	...	...	...	...	...	...	...	...	...	...
891e2055bc3ffff	-0.499624	-0.015107	0.130929	-0.192005	0.520919	-0.350832	0.242332	-0.776085	-0.352089	0.773911	...	-0.584363	-0.091274	-0.211894	-0.304317	0.270350	0.190800	0.388361	0.321336	0.761356	2
891e2055bc7ffff	-0.353990	0.102986	0.096897	-0.284011	0.792185	-0.537846	0.339987	-0.508364	-0.397602	0.779856	...	-0.611381	-0.034055	-0.250024	-0.416371	0.289269	0.047325	0.326037	0.210747	0.463528	2
891e2055bcbffff	-0.649712	-0.050774	0.189476	-0.130992	0.493045	-0.208345	0.289957	-1.025107	-0.309696	0.747634	...	-0.514004	-0.267735	-0.272899	-0.313647	0.365553	0.275041	0.450990	0.498539	1.033792	2
891e205a967ffff	-0.605251	0.114015	0.175078	-0.281711	0.336612	0.154542	-0.026165	-0.365951	-0.396474	0.531186	...	-0.325677	0.356714	0.028953	0.215758	-0.004623	0.233533	0.279212	0.088734	0.554686	1
891e205a9a7ffff	-1.404263	-1.382761	0.503229	-1.089563	0.939539	-0.933452	0.222206	-0.786773	-0.500759	1.735679	...	-0.589892	-0.091554	0.035516	-0.432250	0.239600	0.643934	1.037692	0.163673	0.953123	2

2034 rows × 31 columns

In [8]:

Copied!

plot_numeric_data(regions_gdf, "cluster", embeddings)
plot_numeric_data(regions_gdf, "cluster", embeddings)

Out[8]:

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