Geospatial Data Pack for Visualization 🗺️

Learn Geographic Mapping with Altair, Vega-Lite and Vega using Curated Datasets

Complete geographic and geophysical data collection for mapping and visualization. This consolidation includes 18 complementary datasets used by 31+ Vega, Vega-Lite, and Altair examples 📊. Perfect for learning geographic visualization techniques including projections, choropleths, point maps, vector fields, and interactive displays.

Source data lives on GitHub and can also be accessed via CDN. The vega-datasets project serves as a common repository for example datasets used across these visualization libraries and related projects.

Why Use This Dataset? 🤔

• Comprehensive Geospatial Types: Explore a variety of core geospatial data models:
  • Vector Data: Includes points (like airports.csv), lines (like londonTubeLines.json), and polygons (like us-10m.json).
  • Raster-like Data: Work with gridded datasets (like windvectors.csv, annual-precip.json).
• Diverse Formats: Gain experience with standard and efficient geospatial formats like GeoJSON (see Table 1, 2, 4), compressed TopoJSON (see Table 1), and plain CSV/TSV (see Table 2, 3, 4) for point data and attribute tables ready for joining.
• Multi-Scale Coverage: Practice visualization across different geographic scales, from global and national (Table 1, 4) down to the city level (Table 1).
• Rich Thematic Mapping: Includes multiple datasets (Table 3) specifically designed for joining attributes to geographic boundaries (like states or counties from Table 1) to create insightful choropleth maps.
• Ready-to-Use & Example-Driven: Cleaned datasets tightly integrated with 31+ official examples (see Appendix) from Altair, Vega-Lite, and Vega, allowing you to immediately practice techniques like projections, point maps, network maps, and interactive displays.
• Python Friendly: Works seamlessly with essential Python libraries like Altair (which can directly read TopoJSON/GeoJSON), Pandas, and GeoPandas, fitting perfectly into the Kaggle notebook environment.

Table of Contents

• Why Use This Dataset?
• Dataset Inventory
• Usage Guide
• Appendix: Geospatial Examples by Technique
• Licenses
• Sources
• Learning Resources

Dataset Inventory 🗂️

This pack includes 18 datasets covering base maps, reference points, statistical data for choropleths, and geophysical data.

1. BASE MAP BOUNDARIES (Topological Data)

2. GEOGRAPHIC REFERENCE POINTS (Point Data) 📍

The choropleth map is a device used for the display of socioeconomic data associated with an areal partition of geographic space. Cartographers emphasize the need to standardize any raw count data by an area-based total before displaying the data in a choropleth map. The standardization process converts the raw data from an absolute measure into a relative measure. However, there is recognition that the standardizing process does not enable the map reader to distinguish between low–low and high–high numerator/denominator differences. This research uses concentration-based classification schemes using Lorenz curves to address some of these issues. A test data set of nonwhite birth rate by county in North Carolina is used to demonstrate how this approach differs from traditional mean–variance-based systems such as the Jenks’ optimal classification scheme.

This dataset contains geometry data for the countries of the world together with their names and country codes in various formats. The primary use case is choropleths, color-coded maps. The data can be read as a pandas DataFrame with geopandas and plotted with matplotlib. See the starter notebook for an example how to do it.

The data was created by Natural Earth. It is in public domain and free to use for any purpose at the time of this writing; you might want to check their Terms of Use.

Photo by KOBU Agency on Unsplash

These are the results obtained from an empirical test looking at the communicative effectiveness between two types of two dimensional (2D) map formats (Choropleth maps, and Cartograms) of the Greater London area of the United Kingdom. Participants were interviewed and observed individually during the procedure. The results contain the recorded measurements of spatial accuracy, and the time taken for each participant to answers 3 test questions. A post-test qualitative reaction of each participants' preference between the two map types is recorded, along with their gender, age, visual impediments, and self-assessed map reading ability.

A comprehensive dataset of 1,513 Pakistani cities, towns, tehsils, districts and places with latitude/longitude, administrative region, population (when available) and Wikidata IDs — ideal for mapping, geospatial analysis, enrichment, and location-based ML.

Why this dataset is valuable:

• Full geocoordinates for every entry (100% coverage) — ready for mapping and spatial joins.
• Wide geographic coverage across all 7 major regions of Pakistan (provinces / administrative regions).
• Wikidata IDs included for reliable cross-referencing and automatic enrichment from external knowledge bases.
• Useful for data scientists, GIS engineers, civic tech projects, academic research, and startups building Pakistan-focused location services.

Highlights (fetched from the data):

• Total rows: 1,513
• Unique places (city field): 1,497
• Rows with population > 0: 526 (≈34.8%)
• Coordinate coverage: 1513 / 1513 (100%) — directly usable with mapping libraries.

Column definitions (short):

• id — Internal numeric row id (unique integer).
• wikiDataId — Wikidata QID (e.g., Q####) for the place; use to fetch rich metadata.
• type — Administrative/place type (e.g., ADM1, ADM2, city, district, tehsil).
• city — Common/local city/place name (short label).
• name — Full name / official name of the place (may include “District”, “Tehsil”, etc.).
• country — Country name (Pakistan).
• countryCode — ISO country code (e.g., PK).
• region — Primary administrative region / province (e.g., Punjab, Sindh).
• regionCode — Short code for region (e.g., PB, KP depending on your encoding).
• regionWdId — Wikidata QID for the region.
• latitude — Latitude in decimal degrees (float).
• longitude — Longitude in decimal degrees (float).
• population — Integer population (0 or NA where unknown).

Typical & high-value use cases:

• Mapping & visualization: choropleth maps, point overlays, heatmaps of population or density.
• Geospatial analysis: distance calculations, nearest-neighbor queries, clustering of urban centers.
• Data enrichment: join with other datasets (OpenStreetMap, Wikidata, census data) using wikiDataId and coordinates.
• Machine learning & NLP: training geolocation models, geoparsing, toponym resolution, place name disambiguation.
• Urban planning & research: analyze distribution of population-ready places vs administrative units.
• Mobile / location-based apps: lookup & reverse geocoding fallback, seeding POI databases for Pakistan.
• Humanitarian & disaster response: baseline location lists for logistics and situational awareness.

Folium makes it easy to visualize data that’s been manipulated in Python on an interactive leaflet map. It enables both the binding of data to a map for choropleth visualizations as well as passing rich vector/raster/HTML visualizations as markers on the map. These files can be used to mark the state boundaries on the map of INDIA using folium library and the CSV also contains the state data and how to use it in our notebooks. I have used it in one of my kernels which can be viewed.

The library has a number of built-in tilesets from OpenStreetMap, Mapbox, and Stamen, and supports custom tilesets with Mapbox or Cloudmade API keys. folium supports both Image, Video, GeoJSON, and TopoJSON overlays. Due to extensible functionalities I find folium the best map plotting library in python. Do give it a try and use it in your kernels.

CrimeMapTutorial is a step-by-step tutorial for learning crime mapping using ArcView GIS or MapInfo Professional GIS. It was designed to give users a thorough introduction to most of the knowledge and skills needed to produce daily maps and spatial data queries that uniformed officers and detectives find valuable for crime prevention and enforcement. The tutorials can be used either for self-learning or in a laboratory setting. The geographic information system (GIS) and police data were supplied by the Rochester, New York, Police Department. For each mapping software package, there are three PDF tutorial workbooks and one WinZip archive containing sample data and maps. Workbook 1 was designed for GIS users who want to learn how to use a crime-mapping GIS and how to generate maps and data queries. Workbook 2 was created to assist data preparers in processing police data for use in a GIS. This includes address-matching of police incidents to place them on pin maps and aggregating crime counts by areas (like car beats) to produce area or choropleth maps. Workbook 3 was designed for map makers who want to learn how to construct useful crime maps, given police data that have already been address-matched and preprocessed by data preparers. It is estimated that the three tutorials take approximately six hours to complete in total, including exercises.

This dataset underlies a choropleth map of Boston area communities in which areas are shaded according to the percentage of the population that was foreign-born during each decade. The data was drawn from the US Census of Population, as well as the American Community Survey.

Synthetic populations for regions of the World (SPW) | Jordan

Dataset information

A synthetic population of a region as provided here, captures the people of the region with selected demographic attributes, their organization into households, their assigned activities for a day, the locations where the activities take place and thus where interactions among population members happen (e.g., spread of epidemics).

License

CC-BY-4.0

Acknowledgment

This project was supported by the National Science Foundation under the NSF RAPID: COVID-19 Response Support: Building Synthetic Multi-scale Networks (PI: Madhav Marathe, Co-PIs: Henning Mortveit, Srinivasan Venkatramanan; Fund Number: OAC-2027541).

Contact information

Henning.Mortveit@virginia.edu

Identifiers

Statistics

Sources

Files description

Base data files (jor_data_v_0_9.zip)

Derived data files

Validation and measures files

OverviewThe multiple hazard index for the United States Counties was designed to map natural hazard relating to exposure to multiple natural disasters. The index was created to provide communities and public health officials with an overview of the risks that are prominent in their county, and to facilitate the comparison of hazard level between counties. Most existing hazard maps focus on a single disaster type. By creating a measure that aggregates the hazard from individual disasters, the increased hazard that results from exposure to multiple natural disasters can be better understood. The multiple hazard index represents the aggregate of hazard from eleven individual disasters. Layers displaying the hazard from each individual disaster are also included.

The hazard index is displayed visually as a choropleth map, with the color blue representing areas with less hazard and red representing areas with higher hazard. Users can click on each county to view its hazard index value, and the level of hazard for each individual disaster. Layers describing the relative level of hazard from each individual disaster are also available as choropleth maps with red areas representing high, orange representing medium, and yellow representing low levels of hazard.Methodology and Data CitationsMultiple Hazard Index

The multiple hazard index was created by coding the individual hazard classifications and summing the coded values for each United States County. Each individual hazard is weighted equally in the multiple hazard index. Alaska and Hawaii were excluded from analysis because one third of individual hazard datasets only describe the coterminous United States.

Avalanche Hazard

University of South Carolina Hazards and Vulnerability Research Institute. “Spatial Hazard Events and Losses Database”. United States Counties. “Avalanches United States 2001-2009”. < http://hvri.geog.sc.edu/SHELDUS/

Downloaded 06/2016.

Classification

Avalanche hazard was classified by dividing counties based upon the number of avalanches they experienced over the nine year period in the dataset. Avalanche hazard was not normalized by total county area because it caused an over-emphasis on small counties, and because avalanches are a highly local hazard.

None = 0 AvalanchesLow = 1 AvalancheMedium = 2-5 AvalanchesHigh = 6-10 Avalanches

Earthquake Hazard

United States Geological Survey. “Earthquake Hazard Maps”. 1:2,000,000. “Peak Ground Acceleration 2% in 50 Years”. < http://earthquake.usgs.gov/hazards/products/conterminous/

. Downloaded 07/2016.

Classification

Peak ground acceleration (% gravity) with a 2% likelihood in 50 years was averaged by United States County, and the earthquake hazard of counties was classified based upon this average.

Low = 0 - 14.25 % gravity peak ground accelerationMedium = 14.26 - 47.5 % gravity peak ground accelerationHigh = 47.5+ % gravity peak ground acceleration

Flood Hazard

United States Federal Emergency Management Administration. “National Flood Hazard Layer”. 1:10,000. “0.2 Percent Annual Flood Area”. < https://data.femadata.com/FIMA/Risk_MAP/NFHL/

. Downloaded 07/2016.

Classification

The National Flood Hazard Layer 0.2 Percent Annual Flood Area was spatially intersected with the United States Counties layer, splitting flood areas by county and adding county information to flood areas. Flood area was aggregated by county, expressed as a fraction of the total county land area, and flood hazard was classified based upon percentage of land that is susceptible to flooding. National Flood Hazard Layer does not cover the entire United States; coverage is focused on populated areas. Areas not included in National Flood Hazard Layer were assigned flood risk of Low in order to include these areas in further analysis.

Low = 0-.001% area susceptibleMedium = .00101 % - .005 % area susceptibleHigh = .00501+ % area susceptible

Heat Wave Hazard

United States Center for Disease Control and Prevention. “National Climate Assessment”. Contiguous United States Counties. “Extreme Heat Events: Heat Wave Days in May - September for years 1981-2010”. Downloaded 06/2016.

Classification

Heat wave was classified by dividing counties based upon the number of heat wave days they experienced over the 30 year time period described in the dataset.

Low = 126 - 171 Heat wave DaysMedium = 172 – 187 Heat wave DaysHigh = 188 – 255 Heat wave Days

Hurricane Hazard

National Oceanic and Atmospheric Administration. Coastal Services Center. “Historical North Atlantic Tropical Cyclone Tracks, 1851-2004”. 1: 2,000,000. < https://catalog.data.gov/dataset/historical-north-atlantic-tropical-cyclone-tracks-1851-2004-direct-download

. Downloaded 06/2016.

National Oceanic and Atmospheric Administration. Coastal Services Center. “Historical North Pacific Tropical Cyclone Tracks, 1851-2004”. 1: 2,000,000. < https://catalog.data.gov/dataset/historical-north-atlantic-tropical-cyclone-tracks-1851-2004-direct-download

. Downloaded 06/2016.

Classification

Atlantic and Pacific datasets were merged. Tropical storm and disturbance tracks were filtered out leaving hurricane tracks. Each hurricane track was assigned the value of the category number that describes that event. Weighting each event by intensity ensures that areas with higher intensity events are characterized as being more hazardous. Values describing each hurricane event were aggregated by United States County, normalized by total county area, and the hurricane hazard of counties was classified based upon the normalized value.

Landslide Hazard

United States Geological Survey. “Landslide Overview Map of the United States”. 1:4,000,000. “Landslide Incidence and Susceptibility in the Conterminous United States”. < https://catalog.data.gov/dataset/landslide-incidence-and-susceptibility-in-the-conterminous-united-states-direct-download

. Downloaded 07/2016.

Classification

The classifications of High, Moderate, and Low landslide susceptibility and incidence from the study were numerically coded, the average value was computed for each county, and the landslide hazard was classified based upon the average value.

Long-Term Drought Hazard

United States Drought Monitor, Drought Mitigation Center, United States Department of Agriculture, National Oceanic and Atmospheric Administration. “Drought Monitor Summary Map”. “Long-Term Drought Impact”. < http://droughtmonitor.unl.edu/MapsAndData/GISData.aspx >. Downloaded 06/2016.

Classification

Short-term drought areas were filtered from the data; leaving only long-term drought areas. United States Counties were assigned the average U.S. Drought Monitor Classification Scheme Drought Severity Classification value that characterizes the county area. County long-term drought hazard was classified based upon average Drought Severity Classification value.

Low = 1 – 1.75 average Drought Severity Classification valueMedium = 1.76 -3.0 average Drought Severity Classification valueHigh = 3.0+ average Drought Severity Classification value

Snowfall Hazard

United States National Oceanic and Atmospheric Administration. “1981-2010 U.S. Climate Normals”. 1: 2,000,000. “Annual Snow Normal”. < http://www1.ncdc.noaa.gov/pub/data/normals/1981-2010/products/precipitation/

. Downloaded 08/2016.

Classification

Average yearly snowfall was joined with point location of weather measurement stations, and stations without valid snowfall measurements were filtered out (leaving 6233 stations). Snowfall was interpolated using least squared distance interpolation to create a .05 degree raster describing an estimate of yearly snowfall for the United States. The average yearly snowfall raster was aggregated by county to yield the average yearly snowfall per United States County. The snowfall risk of counties was classified by average snowfall.

None = 0 inchesLow = .01- 10 inchesMedium = 10.01- 50 inchesHigh = 50.01+ inches

Tornado Hazard

United States National Oceanic and Atmospheric Administration Storm Prediction Center. “Severe Thunderstorm Database and Storm Data Publication”. 1: 2,000,000. “United States Tornado Touchdown Points 1950-2004”. < https://catalog.data.gov/dataset/united-states-tornado-touchdown-points-1950-2004-direct-download

. Downloaded 07/2016.

Classification

Each tornado touchdown point was assigned the value of the Fujita Scale that describes that event. Weighting each event by intensity ensures that areas with higher intensity events are characterized as more hazardous. Values describing each tornado event were aggregated by United States County, normalized by total county area, and the tornado hazard of counties was classified based upon the normalized value.

Volcano Hazard

Smithsonian Institution National Volcanism Program. “Volcanoes of the World”. “Holocene Volcanoes”. < http://volcano.si.edu/search_volcano.cfm

. Downloaded 07/2016.

Classification

Volcano coordinate locations from spreadsheet were mapped and aggregated by United States County. Volcano count was normalized by county area, and the volcano hazard of counties was classified based upon the number of volcanoes present per unit area.

None = 0 volcanoes/100 kilometersLow = 0.000915 - 0.007611 volcanoes / 100 kilometersMedium = 0.007612 - 0.018376 volcanoes / 100 kilometersHigh = 0.018377- 0.150538 volcanoes / 100 kilometers

Wildfire Hazard

United States Department of Agriculture, Forest Service, Fire, Fuel, and Smoke Science Program. “Classified 2014 Wildfire Hazard Potential”. 270 meters. < http://www.firelab.org/document/classified-2014-whp-gis-data-and-maps

. Downloaded 06/2016.

Classification

The classifications of Very High, High, Moderate, Low, Very Low, and Non-Burnable/Water wildfire hazard from the study were numerically coded, the average value was computed for each county, and the wildfire hazard was classified based upon the average value.

Synthetic populations for regions of the World (SPW) | Delhi

Dataset information

A synthetic population of a region as provided here, captures the people of the region with selected demographic attributes, their organization into households, their assigned activities for a day, the locations where the activities take place and thus where interactions among population members happen (e.g., spread of epidemics).

License

CC-BY-4.0

Acknowledgment

This project was supported by the National Science Foundation under the NSF RAPID: COVID-19 Response Support: Building Synthetic Multi-scale Networks (PI: Madhav Marathe, Co-PIs: Henning Mortveit, Srinivasan Venkatramanan; Fund Number: OAC-2027541).

Contact information

Henning.Mortveit@virginia.edu

Identifiers

Statistics

Sources

Files description

Base data files (ind_140001944_data_v_0_9.zip)

Derived data files

Validation and measures files

PostGIS data for London and Greater London ward boundaries as of 2018.

This dataset is used in the london_votes sample Splitfile in which the 2017 General Election results and London Ward geodata are joined through the ONS UK Ward-Constituency lookup table to build a dataset of London constituencies and Conservative/Labour votes in each, ready for plotting as a Choropleth map.

https://data.london.gov.uk/dataset/statistical-gis-boundary-files-london

Contains National Statistics data © Crown copyright and database right 2012

Contains Ordnance Survey data © Crown copyright and database right 2012

Synthetic populations for regions of the World (SPW) | Manipur

Dataset information

A synthetic population of a region as provided here, captures the people of the region with selected demographic attributes, their organization into households, their assigned activities for a day, the locations where the activities take place and thus where interactions among population members happen (e.g., spread of epidemics).

License

CC-BY-4.0

Acknowledgment

This project was supported by the National Science Foundation under the NSF RAPID: COVID-19 Response Support: Building Synthetic Multi-scale Networks (PI: Madhav Marathe, Co-PIs: Henning Mortveit, Srinivasan Venkatramanan; Fund Number: OAC-2027541).

Contact information

Henning.Mortveit@virginia.edu

Identifiers

Statistics

Sources

Files description

Base data files (ind_140001942_data_v_0_9.zip)

Derived data files

Validation and measures files

Dataset info

The dataset contains Local Authority Boundaries for Great Britain (England, Scotland and Wales) as of December 2021. A total of 363 Local Authority objects are included. Created for future use in folium choropleth maps when combined with other datasets that contain the matching Local Authority Codes. Additionally, subsets were created for convenience holding the boundaries of local authorities in England and Wales together, and in each individual country, i.e., England, Scotland and Wales on their own.

Methodology

The original dataset was downloaded from ONS. Since the dataset was too large for most use cases (129.4MB) due to the level of detail, it was simplified with https://mapshaper.org/ using the default method (Visvalingam / weighted area) with 'prevent shape removal' enabled. The simplification was set to 1.4%, followed by intersection repair and export back to geojson. The shape coordinates were originally in British National Grid (BNG) format, which had to be converted to WGS84 (latitude and longitude) format. Finally, the coordinates were rounded to 6 decimal places, resulting in a file containing 2.2MB of uncompressed data with a sensible level of detail. The individual country data were extracted, based on the LAD21CD property, to create the additional files.

Licence

https://www.ons.gov.uk/methodology/geography/licences

Digital boundary products and reference maps are supplied under the Open Government Licence. You must use the following copyright statements when you reproduce or use this material:

• Source: Office for National Statistics licensed under the Open Government Licence v.3.0
• Contains OS data © Crown copyright and database right 2023

These are 2 files with Geographic vector layers with Continent Boundaries. You can use this data to create Choropleth maps or for any map visualization that requires vectors at the Continent level.

Citation of the data source:

Shepherd, Stephanie (2020). Continent Polygons. figshare. Dataset. https://doi.org/10.6084/m9.figshare.12555170.v3

I transformed the data lightly to make a GPKG file and to add the documents to Kaggle.

The original data is similar to the file continent_boundaries_8.gpkg, which includes the following continents:

• Africa
• Antarctica
• Asia
• Australia
• Europe
• North America
• Oceania
• South America

I also created a file called continent_boundaries_7.gpkg that merges Australia and Oceania as a single continent.

You can find the transformations in the following Kaggle Notebook: https://www.kaggle.com/code/ericnarro/create-continents-geodataframe-and-file/notebook

This is dataset which you can find population of Ecuadorian cities in 2022 . The data downloaded from this website. In my case, I utilize this data for making choropleth map for analyzing data of "Store Sales - Time Series Forecasting" data and please freely utilize this data for such use. (Thank you very much for "World Population Review"!)

Context

The dataset consists of COVID-19 cases in Malaysia from 27 March 2020 to 15 April 2021. This dataset is collected for the purpose of creating better visualizations for the COVID-19 cases in Malaysia. All of the data is web scraped from https://kpkesihatan.com/ by using BeautifulSoup library.

The data is also available in GitHub, along with the scripts made to scrape the data. There is also a Web Application made to show the visualizations.

Originally I planned to update the data daily but I find that it seems too tedious for me to do this alone without some sort of automated scripts or schedulers. I have been wondering how to do this efficiently with automation or schedulers, if someone knows how to do this efficiently, please reach out to me by emailing or message in LinkedIn, the links can be found in my GitHub, thank you very much.

Content

There are three CSV files and one GeoJSON file: - all_2020-03-27_2021-04-15.csv: all daily cases excluding state data - state_all.csv: all daily cases for each state - state_cumu.csv: all daily cumulative cases for each state - malaysia_state_province_boundary.geojson: Malaysia's GeoJSON map file

The columns consist of: 1. Date 2. Recovered 3. Cumulative Recovered 4. Imported Case (many NaN values till the end of 2020) 5. Local Case (many NaN values) 6. Active Case (many NaN values but can be inferred) 7. New Case 8. Cumulative Case 9. ICU - Number of patients admitted into Intensive Care Unit 10. Ventilator - Number of patients who need ventilator in ICU 11. Death 12. Cumulative Death 13. URL - link to the original webpage

Acknowledgements

Thanks to Info GIS MAP.com that provides Malaysia's GeoJSON file to create Choropleth maps.

Inspiration

Hopefully, there will be people utilizing the scripts or the data to create better visualizations.

How was the data collected? The dataset was collected from Crop Residue Burning Information and Management System (CRBIMS), Punjab Remote Sensing Centre (PRSC). The data is collected for all the districts of Punjab from April 2016 to June 2024.

What can be done with this data?

Visualization: This data can be used for visualization purposes. Choropleth maps can be made using the coordinates available.

https://github.com/waitasecant/CRBIMS/blob/main/trends.png?raw=true" alt="">

Source: http://gis-prsc.punjab.gov.in/residue/Index.aspx

About Datasets

Tokyo is the largest prefecture and has the largest number of cases of COVID-19 in Japan. The number of total confirmed cases in Tokyo is about 73000 (as of January 9th, 2021). In this dataset, data about COVID-19 in Tokyo contain. If you want to download it, please consider upvoting.

Data Source

Data was collected from Tokyo Metropolitan Government Open Data Catalog Site and Updates on COVID-19 in Tokyo.

Columns

tokyo_covid19_patients.csv file in this dataset has 7 columns. | Column | Description | | --- | --- | | Number | | | Date | Published Date | | Date (Onset) | Date of onset of symptoms | | Region | Region where patients live in | | Age | Patients age| | Gender | Patients gender| | Situation | This columns shows whether the patient was discharged (include death) or not.|

tokyo_cases_byarea.csv has 4 columns. | Column | Description | | --- | --- | | Area | This column shows that which area the municipality belong. | | Municipality | Municipality name | | Positive Cases | The number of total cases | | Code | Code required to draw a choropleth map |