Overture Maps Loader¶

OvertureMapsLoader can download the Overture Maps data from the s3 bucket for a given region.

It is a wrapper around the OvertureMaestro library that can download the data in the original format but also have some advanced functions.

In the SRAI context, OvertureMapsLoader utilizes so-called wide format for returning features with columns representing potential categories of the object. If you want to read more in-depth about this format, you can checkout this OvertureMaestro's docs page.

Run this notebook in Google Colab:

Remember to install the srai library before running the notebook:

%pip install srai[all]

In [1]:

Copied!

from shapely.geometry import box

from srai.constants import GEOMETRY_COLUMN
from srai.loaders import OvertureMapsLoader
from srai.regionalizers import geocode_to_region_gdf
from shapely.geometry import box

from srai.constants import GEOMETRY_COLUMN
from srai.loaders import OvertureMapsLoader
from srai.regionalizers import geocode_to_region_gdf

Using OvertureMapsLoader to download data for a specific area¶

Download all available features in Paris, France¶

In [2]:

Copied!





loader = OvertureMapsLoader()
paris = geocode_to_region_gdf("Paris")
paris_features_gdf = loader.load(paris)
paris_features_gdf
loader = OvertureMapsLoader()
paris = geocode_to_region_gdf("Paris")
paris_features_gdf = loader.load(paris)
paris_features_gdf

Finished operation in 0:00:26

Out[2]:

	geometry	base\|infrastructure\|aerialway	base\|infrastructure\|airport	base\|infrastructure\|barrier	base\|infrastructure\|bridge	base\|infrastructure\|communication	base\|infrastructure\|emergency	base\|infrastructure\|manhole	base\|infrastructure\|pedestrian	base\|infrastructure\|pier	...	places\|place\|professional_services	places\|place\|public_service_and_government	places\|place\|real_estate	places\|place\|religious_organization	places\|place\|retail	places\|place\|structure_and_geography	places\|place\|travel	transportation\|segment\|rail	transportation\|segment\|road	transportation\|segment\|water
feature_id
08b1fb46666c3fff0001a862ac62e27e	POINT (2.34731 48.86724)	False	False	True	False	False	False	False	False	False	...	False	False	False	False	False	False	False	False	False	False
08b1fb46666cefff0001ad93abc79405	POINT (2.34739 48.86729)	False	False	True	False	False	False	False	False	False	...	False	False	False	False	False	False	False	False	False	False
08b1fb46666c1fff0001a39bc9a68810	POINT (2.3473 48.8673)	False	False	True	False	False	False	False	False	False	...	False	False	False	False	False	False	False	False	False	False
08b1fb46666c1fff0001aa11990adc4d	POINT (2.34741 48.86733)	False	False	False	False	False	False	False	False	False	...	False	False	False	False	False	False	False	False	False	False
08b1fb46666c1fff0001ad131e8b11cc	POINT (2.34717 48.86734)	False	False	True	False	False	False	False	False	False	...	False	False	False	False	False	False	False	False	False	False
...	...	...	...	...	...	...	...	...	...	...	...	...	...	...	...	...	...	...	...	...	...
08b1fb4759a49fff020097d3e602ce58	POLYGON ((2.33595 48.90143, 2.33573 48.90155, ...	False	False	False	False	False	False	False	False	False	...	False	False	False	False	False	False	False	False	False	False
08b1fb4759a49fff02007817c242243d	POLYGON ((2.3362 48.9014, 2.33597 48.90153, 2....	False	False	False	False	False	False	False	False	False	...	False	False	False	False	False	False	False	False	False	False
08b1fb4664d11fff0200a4fea2da5f4b	POLYGON ((2.3375 48.90124, 2.33756 48.9012, 2....	False	False	False	False	False	False	False	False	False	...	False	False	False	False	False	False	False	False	False	False
08b1fb4664d1cfff020073b742aa7b3b	POLYGON ((2.33818 48.90131, 2.3381 48.90135, 2...	False	False	False	False	False	False	False	False	False	...	False	False	False	False	False	False	False	False	False	False
08b1fb4664d03fff02009c2aabdb5591	POLYGON ((2.33841 48.90134, 2.3384 48.90134, 2...	False	False	False	False	False	False	False	False	False	...	False	False	False	False	False	False	False	False	False	False

570790 rows × 115 columns

Plot features¶

Colours from the this palette: https://colorhunt.co/palette/f8ededff8225b43f3f173b45

In [3]:

Copied!





ax = paris.plot(color="#F8EDED", figsize=(16, 16))

# plot water
water_columns = [c for c in paris_features_gdf.columns if "water" in c]
water_data = paris_features_gdf[paris_features_gdf[water_columns].any(axis=1)]
water_data.plot(ax=ax, color="#FF8225", markersize=0)

# plot_roads
roads_data = paris_features_gdf[paris_features_gdf["transportation|segment|road"]]
roads_data.plot(ax=ax, color="#B43F3F", markersize=0, linewidth=0.25)

# plot buildings
building_columns = [c for c in paris_features_gdf.columns if c.startswith("buildings")]
buildings_data = paris_features_gdf[paris_features_gdf[building_columns].any(axis=1)]
buildings_data.plot(ax=ax, color="#173B45", markersize=0)

paris.boundary.plot(ax=ax, color="#173B45", linewidth=2, alpha=0.5)

xmin, ymin, xmax, ymax = paris.total_bounds
ax.set_xlim(xmin - 0.001, xmax + 0.001)
ax.set_ylim(ymin - 0.001, ymax + 0.001)

ax.set_axis_off()
ax = paris.plot(color="#F8EDED", figsize=(16, 16))

# plot water
water_columns = [c for c in paris_features_gdf.columns if "water" in c]
water_data = paris_features_gdf[paris_features_gdf[water_columns].any(axis=1)]
water_data.plot(ax=ax, color="#FF8225", markersize=0)

# plot_roads
roads_data = paris_features_gdf[paris_features_gdf["transportation|segment|road"]]
roads_data.plot(ax=ax, color="#B43F3F", markersize=0, linewidth=0.25)

# plot buildings
building_columns = [c for c in paris_features_gdf.columns if c.startswith("buildings")]
buildings_data = paris_features_gdf[paris_features_gdf[building_columns].any(axis=1)]
buildings_data.plot(ax=ax, color="#173B45", markersize=0)

paris.boundary.plot(ax=ax, color="#173B45", linewidth=2, alpha=0.5)

xmin, ymin, xmax, ymax = paris.total_bounds
ax.set_xlim(xmin - 0.001, xmax + 0.001)
ax.set_ylim(ymin - 0.001, ymax + 0.001)

ax.set_axis_off()

No description has been provided for this image

Download more detailed data with higher hierarchy value¶

By default, the hierarchy_depth value is equal to 1, but it can be set to None to get a full list of all possible columns.

In [4]:

Copied!





manhattan_bbox = box(-73.994551, 40.762396, -73.936872, 40.804239)
loader = OvertureMapsLoader(hierarchy_depth=None)
new_york_features_gdf = loader.load(manhattan_bbox)
new_york_features_gdf
manhattan_bbox = box(-73.994551, 40.762396, -73.936872, 40.804239)
loader = OvertureMapsLoader(hierarchy_depth=None)
new_york_features_gdf = loader.load(manhattan_bbox)
new_york_features_gdf

Finished operation in 0:00:31

Out[4]:

	geometry	base\|infrastructure\|aerialway\|aerialway_station	base\|infrastructure\|aerialway\|cable_car	base\|infrastructure\|aerialway\|chair_lift	base\|infrastructure\|aerialway\|drag_lift	base\|infrastructure\|aerialway\|gondola	base\|infrastructure\|aerialway\|goods	base\|infrastructure\|aerialway\|j-bar	base\|infrastructure\|aerialway\|magic_carpet	base\|infrastructure\|aerialway\|mixed_lift	...	transportation\|segment\|road\|service\|parking_aisle	transportation\|segment\|road\|steps	transportation\|segment\|road\|tertiary	transportation\|segment\|road\|tertiary\|link	transportation\|segment\|road\|track	transportation\|segment\|road\|trunk	transportation\|segment\|road\|trunk\|link	transportation\|segment\|road\|unclassified	transportation\|segment\|road\|unknown	transportation\|segment\|water
feature_id
08b2a100d2ce9fff0001bbf87e990781	LINESTRING (-73.97214 40.7268, -73.97341 40.72...	False	False	False	False	False	False	False	False	False	...	False	False	False	False	False	False	False	False	False	False
08b2a10725843fff0001aed186cf90e1	POINT (-73.99301 40.7627)	False	False	False	False	False	False	False	False	False	...	False	False	False	False	False	False	False	False	False	False
08b2a10725850fff0001a8078eb60797	POINT (-73.99422 40.76333)	False	False	False	False	False	False	False	False	False	...	False	False	False	False	False	False	False	False	False	False
08b2a10725850fff0001abb9a995714a	POINT (-73.99409 40.76328)	False	False	False	False	False	False	False	False	False	...	False	False	False	False	False	False	False	False	False	False
08b2a10725842fff0001abd29d53f39b	POINT (-73.9933 40.76271)	False	False	False	False	False	False	False	False	False	...	False	False	False	False	False	False	False	False	False	False
...	...	...	...	...	...	...	...	...	...	...	...	...	...	...	...	...	...	...	...	...	...
08b2a100f2650fff02008c2684f54385	POLYGON ((-73.93664 40.77279, -73.93679 40.772...	False	False	False	False	False	False	False	False	False	...	False	False	False	False	False	False	False	False	False	False
08b2a100f2656fff0200f667ed909313	POLYGON ((-73.93718 40.77323, -73.93719 40.773...	False	False	False	False	False	False	False	False	False	...	False	False	False	False	False	False	False	False	False	False
08b2a100f260bfff0200ec356893da2f	POLYGON ((-73.93737 40.77379, -73.93741 40.773...	False	False	False	False	False	False	False	False	False	...	False	False	False	False	False	False	False	False	False	False
08b2a100f2604fff0200b0ba2bc875ff	POLYGON ((-73.93743 40.7752, -73.93661 40.7749...	False	False	False	False	False	False	False	False	False	...	False	False	False	False	False	False	False	False	False	False
08b2a100f2622fff0200df47d5bd8207	POLYGON ((-73.93699 40.77577, -73.93694 40.775...	False	False	False	False	False	False	False	False	False	...	False	False	False	False	False	False	False	False	False	False

64539 rows × 2633 columns

As you can see, there are over 2600 columns available.

Let's see top 20 most popular columns.

In [5]:

Copied!

new_york_features_gdf.drop(columns=GEOMETRY_COLUMN).sum().sort_values(ascending=False).head(20)
new_york_features_gdf.drop(columns=GEOMETRY_COLUMN).sum().sort_values(ascending=False).head(20)

Out[5]:

buildings|building                                                          12487
base|land|tree|tree                                                          5670
base|infrastructure|transportation|crossing                                  3848
transportation|segment|road|footway                                          3379
base|infrastructure|barrier|kerb                                             3025
places|place|health_and_medical                                              2947
places|place|health_and_medical|doctor                                       2278
transportation|segment|road|footway|sidewalk                                 2145
places|place|eat_and_drink|restaurant                                        1742
buildings|building|residential|apartments                                    1706
transportation|segment|road|footway|crosswalk                                1645
places|place|professional_services                                           1355
base|infrastructure|transit|bicycle_parking                                   994
base|infrastructure|transportation|traffic_signals                            990
places|place|retail                                                           867
places|place|beauty_and_spa                                                   853
transportation|segment|road|residential                                       832
places|place|health_and_medical|hospital                                      830
places|place|beauty_and_spa|beauty_salon                                      791
places|place|attractions_and_activities|landmark_and_historical_building      749
dtype: int64

Configure places dataset¶

Places schema is the only one that is treated differently than other data types.

By default, places use both primary and alternate categories to define a feature.

Additionally, there is a filter applied to get only features with confidence score >= 0.75.

There are two dedicated parameters: places_minimal_confidence and places_use_primary_category_only to configure how the data should be transformed.

Let's do example with both of these parameters. We will also use a theme_type_pairs parameter to limit the scope of the downloaded data.

In [6]:

Copied!





default_confidence_loader = OvertureMapsLoader(
    theme_type_pairs=[("places", "place")], places_use_primary_category_only=True
)
strict_confidence_loader = OvertureMapsLoader(
    theme_type_pairs=[("places", "place")],
    places_minimal_confidence=0.99,
    places_use_primary_category_only=True,
)
songpa = geocode_to_region_gdf("Songpa-gu, Seoul")
songpa_default_confidence_features_gdf = default_confidence_loader.load(songpa)
songpa_strict_confidence_features_gdf = strict_confidence_loader.load(songpa)

print(f"Default confidence score: {len(songpa_default_confidence_features_gdf)}")
print(f"Strict confidence score:  {len(songpa_strict_confidence_features_gdf)}")
default_confidence_loader = OvertureMapsLoader(
    theme_type_pairs=[("places", "place")], places_use_primary_category_only=True
)
strict_confidence_loader = OvertureMapsLoader(
    theme_type_pairs=[("places", "place")],
    places_minimal_confidence=0.99,
    places_use_primary_category_only=True,
)
songpa = geocode_to_region_gdf("Songpa-gu, Seoul")
songpa_default_confidence_features_gdf = default_confidence_loader.load(songpa)
songpa_strict_confidence_features_gdf = strict_confidence_loader.load(songpa)

print(f"Default confidence score: {len(songpa_default_confidence_features_gdf)}")
print(f"Strict confidence score:  {len(songpa_strict_confidence_features_gdf)}")

Finished operation in 0:00:03

Finished operation in 0:00:03

Default confidence score: 5224
Strict confidence score:  3

Let's see the count of categories in the places dataset with confidence score >= 0.99.

In [7]:

Copied!





songpa_strict_confidence_features_df = songpa_strict_confidence_features_gdf.drop(
    columns=GEOMETRY_COLUMN
)
songpa_strict_confidence_features_df.sum().loc[lambda x: x > 0].sort_values(ascending=False)
songpa_strict_confidence_features_df = songpa_strict_confidence_features_gdf.drop(
    columns=GEOMETRY_COLUMN
)
songpa_strict_confidence_features_df.sum().loc[lambda x: x > 0].sort_values(ascending=False)

Out[7]:

places|place|retail    3
dtype: int64

Plot features¶

Now we will see the difference between default list of places (gray dots) and strict ones (coloured circles)

In [8]:

Copied!





m = songpa_default_confidence_features_gdf.loc[
    songpa_default_confidence_features_gdf.index.difference(
        songpa_strict_confidence_features_gdf.index
    )
].geometry.explore(
    tiles="CartoDB Voyager",
    color="gray",
    tooltip=False,
    style_kwds=dict(opacity=0.25, stroke=False),
)
songpa.boundary.explore(m=m, color="black")

songpa_gdf_with_categories = songpa_strict_confidence_features_gdf.join(
    songpa_strict_confidence_features_df.dot(songpa_strict_confidence_features_df.columns).rename(
        "category"
    )
)
songpa_gdf_with_categories.geometry.explore(
    m=m,
    tooltip=False,
    marker_kwds=dict(radius=6),
    style_kwds=dict(color="black", fillOpacity=1),
)
songpa_gdf_with_categories[[GEOMETRY_COLUMN, "category"]].explore(
    m=m,
    column="category",
    tooltip=["feature_id", "category"],
    cmap="tab20",
    marker_kwds=dict(radius=4),
    style_kwds=dict(fillOpacity=1),
)
m = songpa_default_confidence_features_gdf.loc[
    songpa_default_confidence_features_gdf.index.difference(
        songpa_strict_confidence_features_gdf.index
    )
].geometry.explore(
    tiles="CartoDB Voyager",
    color="gray",
    tooltip=False,
    style_kwds=dict(opacity=0.25, stroke=False),
)
songpa.boundary.explore(m=m, color="black")

songpa_gdf_with_categories = songpa_strict_confidence_features_gdf.join(
    songpa_strict_confidence_features_df.dot(songpa_strict_confidence_features_df.columns).rename(
        "category"
    )
)
songpa_gdf_with_categories.geometry.explore(
    m=m,
    tooltip=False,
    marker_kwds=dict(radius=6),
    style_kwds=dict(color="black", fillOpacity=1),
)
songpa_gdf_with_categories[[GEOMETRY_COLUMN, "category"]].explore(
    m=m,
    column="category",
    tooltip=["feature_id", "category"],
    cmap="tab20",
    marker_kwds=dict(radius=4),
    style_kwds=dict(fillOpacity=1),
)

Out[8]:

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