Building on our previous guide on installing GeoPandas, we now turn to a recent development: the introduction of the geodatasets package. This new package, separated from the main GeoPandas library, focuses on access to and management of spatial datasets. We’ll walk you through installing geodatasets and demonstrate how it complements GeoPandas in handling geospatial information.
Installing Geodatasets
First, make sure you have GeoPandas installed as per our previous guide. Then, install the geodatasets package in the same environment where you installed GeoPandas:
conda install geodatasets -c conda-forge
It’s worth noting that before the introduction of the geodatasets library, geospatial datasets were often accessed and manipulated without needing a specialized library. Users could download these datasets using geopandas.datasets. Along with the release of GeoPandas Version 1.0.0 on June 24th 2024, this was deprecated and removed. Going forth, new sample datasets are made available via the geodatasets package.
Exploring Available Datasets
Once the new package has been installed, we can begin to see what datasets are available:
import geodatasets
# Get a list of all dataset keys
dataset_keys = list(geodatasets.data.flatten().keys())
# Print the number of available datasets
print(f"Total available datasets: {len(dataset_keys)}")
Output:
Total available datasets: 58
We currently have a total of 58 datasets available within the geodatasets package. Let’s proceed to check out few of them.
print("Some available datasets:")
for key in dataset_keys[:5]: #Adjust the slice to show more or fewer examples
print(f" - {key}")
This will show you the first five datasets available.
Some available datasets: - geoda.airbnb - geoda.atlanta - geoda.cars - geoda.charleston1 - geoda.charleston2
Reviewing a Dataset
Let’s pick a dataset and look at its details:
chosen_dataset = 'geoda.airbnb'
dataset_info = geodatasets.data.query_name(chosen_dataset)
print("Dataset Information:")
print(f"Name: {dataset_info['name']}")
print(f"Description: {dataset_info['description']}")
print(f"Geometry Type: {dataset_info['geometry_type']}")
print(f"Number of Rows: {dataset_info['nrows']}")
This gives us an overview of the dataset:
Dataset Information: Name: geoda.airbnb Description: Airbnb rentals, socioeconomics, and crime in Chicago Geometry Type: Polygon Number of Rows: 77
This information tells us that we’re working with polygon data, likely representing different areas of Chicago, with information about Airbnb rentals, socioeconomic factors, and crime statistics.
Loading and Examining the Data
Now, let’s load the data and take a closer look. We’ll break this down into several steps for easier understanding.
Step 1: Loading the Data
First, we’ll load the dataset and take a look at its columns:
import geopandas as gpd
# Get the path to the dataset
path = geodatasets.get_path('geoda.airbnb')
# Load the dataset
gdf = gpd.read_file(path)
# Display the column names
print("Dataset columns:")
print(gdf.columns)
Output:
Dataset columns:
Index(['community', 'shape_area', 'shape_len', 'AREAID', 'response_r',
'accept_r', 'rev_rating', 'price_pp', 'room_type', 'num_spots',
'poverty', 'crowded', 'dependency', 'without_hs', 'unemployed',
'income_pc', 'harship_in', 'num_crimes', 'num_theft', 'population',
'geometry'],
dtype='object')
This output shows us all the columns in our dataset, including both spatial (geometry) and non-spatial attributes.
Step 2: Examining the Data
Next, let’s look at the first few rows of our data:
# Display the first few rows
print("\nFirst few rows:")
print(gdf.head())
Output:
First few rows:
community shape_area shape_len AREAID response_r \
0 DOUGLAS 46004621.1581 31027.0545098 35 98.771429
1 OAKLAND 16913961.0408 19565.5061533 36 99.200000
2 FULLER PARK 19916704.8692 None 37 68.000000
3 GRAND BOULEVARD 48492503.1554 28196.8371573 38 94.037037
4 KENWOOD 29071741.9283 23325.1679062 39 92.542857
accept_r rev_rating price_pp room_type num_spots ... crowded \
0 94.514286 87.777778 78.157895 1.789474 38 ... 1.8
1 90.105263 88.812500 53.775000 1.850000 20 ... 1.3
2 NaN 91.750000 84.000000 1.833333 6 ... 3.2
3 83.615385 92.750000 119.533333 1.533333 30 ... 3.3
4 88.142857 90.656250 77.991453 1.615385 39 ... 2.4
dependency without_hs unemployed income_pc harship_in num_crimes \
0 30.7 14.3 18.2 23791 47 5013
1 40.4 18.4 28.7 19252 78 1306
2 44.9 26.6 33.9 10432 97 1764
3 39.5 15.9 24.3 23472 57 6416
4 35.4 11.3 15.7 35911 26 2713
num_theft population geometry
0 1241 18238 POLYGON ((-87.609 41.845, -87.609 41.845, -87....
1 311 5918 POLYGON ((-87.592 41.817, -87.592 41.817, -87....
2 383 2876 POLYGON ((-87.629 41.802, -87.629 41.802, -87....
3 1428 21929 POLYGON ((-87.607 41.817, -87.607 41.817, -87....
4 654 17841 POLYGON ((-87.592 41.817, -87.592 41.817, -87....
[5 rows x 21 columns]
This gives us a glimpse of the data, showing various attributes for each community along with their geometric information.
Step 3: Understanding the Data Structure
To get more detailed information about our dataset, we can use the info() method:
# Display basic information about the dataset
print("\nDataset info:")
print(gdf.info())
Output:
Dataset info: <class 'geopandas.geodataframe.GeoDataFrame'> RangeIndex: 77 entries, 0 to 76 Data columns (total 21 columns): # Column Non-Null Count Dtype --- ------ -------------- ----- 0 community 77 non-null object 1 shape_area 77 non-null object 2 shape_len 76 non-null object 3 AREAID 77 non-null int64 4 response_r 68 non-null float64 5 accept_r 67 non-null float64 6 rev_rating 67 non-null float64 7 price_pp 69 non-null float64 8 room_type 69 non-null float64 9 num_spots 77 non-null int64 10 poverty 77 non-null float64 11 crowded 77 non-null float64 12 dependency 77 non-null float64 13 without_hs 77 non-null float64 14 unemployed 77 non-null float64 15 income_pc 77 non-null int64 16 harship_in 77 non-null int64 17 num_crimes 77 non-null int64 18 num_theft 77 non-null int64 19 population 77 non-null int64 20 geometry 77 non-null geometry dtypes: float64(10), geometry(1), int64(7), object(3) memory usage: 12.8+ KB
This output provides a detailed overview of our dataset, including the number of entries, data types of each column, and memory usage.
Step 4: Examining Spatial Components
Finally, let’s look at the spatial aspects of our GeoDataFrame:
# Display the coordinate reference system (CRS)
print("\nCoordinate Reference System:")
print(gdf.crs)
# Display the geometry column's data type
print("\nGeometry column type:")
print(gdf.geometry.dtype)
# Display a summary of the geometry column
print("\nGeometry column summary:")
print(gdf.geometry.head())
Output:
Coordinate Reference System: EPSG:4326 Geometry column type: geometry Geometry column summary: 0 POLYGON ((-87.609 41.845, -87.609 41.845, -87.... 1 POLYGON ((-87.592 41.817, -87.592 41.817, -87.... 2 POLYGON ((-87.629 41.802, -87.629 41.802, -87.... 3 POLYGON ((-87.607 41.817, -87.607 41.817, -87.... 4 POLYGON ((-87.592 41.817, -87.592 41.817, -87.... Name: geometry, dtype: geometry
This final part shows us the spatial components of our GeoDataFrame: the Coordinate Reference System (CRS), the data type of the geometry column, and a summary of the first few geometry entries.
Understanding the Changes
The move from geopandas.datasets to the separate geodatasets package offers several benefits:
- Modularity: It allows for more frequent updates to sample datasets without requiring updates to GeoPandas itself.
- Richer Metadata: Each dataset comes with detailed information about its contents, source, and usage.
- Expanded Collection: The new package offers a wider variety of datasets for different use cases.
Conclusion
While the deprecation of geopandas.datasets might seem disruptive at first, the new geodatasets package offers a more robust and flexible way to access sample geospatial data. By separating the datasets from the main GeoPandas package, the developers have made it easier to maintain and expand the collection of sample data available to users.
This example with the Chicago Airbnb dataset demonstrates how we can use GeoPandas and the new geodatasets package to explore socioeconomic data with a geographical component. We’ve seen how to load the data into a GeoDataFrame and examine its structure, setting the stage for more advanced geospatial analysis and visualization in future posts.
