Published September 20, 2022 | Version 3.0.0
Dataset Open

Swiss Dwellings: A large dataset of apartment models including aggregated geolocation-based simulation results covering viewshed, natural light, traffic noise, centrality and geometric analysis

Description

Introduction

This dataset contains detailed data on over 45,000 apartments (370,000 rooms) in ~3,100 buildings including their geometries, room typology as well as their visual, acoustical, topological, and daylight characteristics. Additionally, we have included location-specific characteristics for the buildings, including climatic data and points of interest within walking distance.

Changelog

  • v3.0.0 (2023-03-31):
    • Updated the dataset increasing the total number of apartments to 45176 and incorporating fixes to some of the sites. The update includes re-digitized apartments and thus alters some ID values.
  • v2.2.1 (2023-03-10):
    • A file, location_ratings.csv, has been included to provide ratings of the locations in which the buildings are situated. The ratings, provided by Fahrländer Partner AG, give insights into the living situation at the buildings' addresses. Details for the different dimensions are provided below.
    • The file location.csv has been updated to include the minimum and maximum temperatures for the locations in which the buildings are situated.
  • v2.1.0 (2022-12-23):
    • A file, locations.csv, has been included to provide information on the climatic and infrastructural characteristics of the locations in which each building is situated
  • v2.0.0 (2022-10-17):
    • Additional to the residential units, we also include the commercial and public parts (such as staircases) of the models. The field unit_usage describes whether an area belongs to a commercial, residential, janitor or public part of the building
    • Added the fields elevation and height to geometries.csv to describe the elevation above the terrain surface and the height of objects.
    • Added the field plan_id which allows identifying which floors are based on the same floor plan (in some cases multiple floors of a building share the same floor plan
    • Improved the ordering of fields in the CSV files (instead of alphabetic order)
    • Minor changes to individual sites

Procurement

The data is sourced from commercial clients of Archilyse AG specializing on the digitization and analysis of buildings. The existing building plans of clients are converted into a geo-referenced, semantically annotated representation and undergo a manual Q/A process to ensure the accuracy of the data and to ensure a maximum 5%-deviation in the apartments' areas (validated with a median deviation of 1.2%).

Geometries

The dataset contains a file geometries.csv which contains the geometries of all areas, walls, railings, columns, windows, doors and features (sinks, bathtubs, etc.) of an apartment.

In total, the datasets contain the 2D geometry of ~1.7 million separators (walls, railings), ~715,000 openings (windows, doors), ca. 520,000 areas (rooms, bathrooms, kitchens, etc.), and ~315,000 features (sinks, toilets, bathtubs, etc.).

Each row contains:

  • apartment_id: The ID of the apartment (for features, areas), note: an apartment id is only unique per site
  • site_id: The ID of the site
  • building_id: The ID of the building
  • floor_id: The ID of the floor
  • plan_id: The ID of the plan on which the floor is based, multiple floors of a building might be based on the same plan
  • unit_id: The ID of the unit in which the element is spatially contained (for features, areas)
  • area_id: The ID of the area in which the element is spatially contained (for features)
  • unit_usage: The usage of the unit, possible values are: RESIDENTIAL, COMMERCIAL, PUBLIC, JANITOR
  • entity_type: The entity type (area, separator, opening, feature)
  • entity_subtype: The entity’s sub-type (e.g. WALL)
  • geometry: The element’s geometry as a WKT geometry in meters. The geometry is given in the site’s local coordinate system. I.e. the position between elements of the same site are correct in respect to each other. The +y direction points northwards, the +x direction points eastwards.
  • elevation: The object's elevation above the terrain surface in meters. We assume one terrain baseline per building, thus all walls in a given floor share the same elevation value. However, windows in particular might start at different elevations and have differing heights.
  • height: The height of the entity in meters, note: In many cases, a default height is assumed

An example:

column  
apartment_id

d4438f2129b30290845ce7eef98a5ba7

site_id 127
building_id 164
plan_id 492
floor_id 861
unit_id 63777
area_id 767676
unit_usage RESIDENTIAL
entity_type area
entity_subtype LIVING_ROOM
geometry

POLYGON ((-6.1501158933490139 -4.8490786654693...

elevation 0
height 2.6

Simulations

Besides the geometrical model, we also provide simulation data on the visual, acoustic, solar, layout, and connectivity-related characteristics of the apartments. The file simulations.csv contains the simulation data aggregated on a per-area basis. Each row contains the identifier columns area_idunit_idapartment_idfloor_idbuilding_idsite_id as defined above as well as 367 simulation columns. Each simulation column is formatted as:

<simulation_category>_<simulation_dimensions>_<aggregation_function>

For instance. the column view_buildings_median describes the amount of building surface that can be seen from any point in a given room. The aggregation methods vary per simulation category and are described in detail below.

Layout

The layout features represent simple features based on the geometry and composition of a room, the dataset provides the following information in an unaggregated form.

Area Basics / Geometry

dimension description
layout_area_type The area’s area type
layout_net_area The area’s share of the apartment’s net area (e.g. 0 for a balcony)
layout_area The area’s actual area
layout_perimeter The area’s perimeter
layout_compactness The area’s compactness (the Polsby–Popper score)
layout_room_count The area’s share to the apartment’s room count
layout_is_navigable True if the area is navigable by a wheelchair

Area Features

dimension description
layout_has_sink True if the area has a sink
layout_has_shower True if the area has a shower
layout_has_bathtub True if the area has a bathtub
layout_has_toilet True if the area has a toilet
layout_has_stairs True if the area has stairs
layout_has_entrance_door True if the area is directly leading to an exit of the apartment

Area Windows / Doors

dimension description
layout_number_of_doors The number of doors directly leading to the area
layout_number_of_windows The number of windows of the area
layout_door_perimeter The sum of all door lengths directly leading to the area
layout_window_perimeter The sum of all window lengths of the area

Area Walls / Railings

dimension description
layout_open_perimeter The sum of all of the boundaries of the area that are neither walls nor railings
layout_railing_perimeter The sum of all of the boundaries of the area that are railings
layout_mean_walllengths The mean length of the area’s sides
layout_std_walllengths The standard deviation of the lengths of the area’s sides

Area Adjacency

dimension description
layout_connects_to_bathroom True if the area connects to a bathroom
layout_connects_to_private_outdoor True if the area connects to an outside area that is private to the apartment

View

The views from an object help to understand the impact of the surroundings on the object. The view simulation calculates the visible amount of buildings, greenery, water, etc. on each individual hexagon from the analyzed object. The values are expressed in steradians (sr) and represent the amount a particular object category occupies in the spherical field of view.

Each of the following dimensions is provided using the room-wise aggregations' minmaxmeanstdmedianp20, and p80. For instance, the column view_greenery_p20 describes the amount of greenery that can be seen from at least 20% of the positions in the area.

dimension description
view_buildings The amount of visible buildings
view_greenery The amount of visible greenery
view_ground The amount of visible ground
view_isovist The amount of visible isovist
view_mountains_class_2 The amount of visible mountains of UN mountain class 2
view_mountains_class_3 The amount of visible mountains of UN mountain class 3
view_mountains_class_4 The amount of visible mountains of UN mountain class 4
view_mountains_class_5 The amount of visible mountains of UN mountain class 5
view_mountains_class_6 The amount of visible mountains of UN mountain class 6
view_railway_tracks The amount of visible railway_tracks
view_site The amount of visible site
view_sky The amount of visible sky
view_tertiary_streets The amount of visible tertiary_streets
view_secondary_streets The amount of visible secondary_streets
view_primary_streets The amount of visible primary_streets
view_pedestrians The amount of visible pedestrians
view_highways The amount of visible highways
view_water The amount of visible water

Sun

Sun simulations help to understand the impact of solar radiation on the object. The outcome of the sun simulations helps to identify surfaces that have great solar potential. Sun simulations are defined by the amount of solar radiation on each individual hexagon from the analyzed object. The sun simulation not only includes direct sun but also considers scattered light. The sun simulation values are given in Kilolux (klx). Simulations are performed for the days of the summer solstice, winter solstice, and the vernal equinox.

Each of the following dimensions is provided using the room-wise aggregations' minmaxmeanstdmedianp20, and p80. For instance, column sun_201806211200_median describes the median amount of direct daylight received on the positions in the area.

Vernal Equinox

dimension description
sun_201803210800 Daylight at 08:00 on 21st of March
sun_201803211000 Daylight at 10:00 on 21st of March
sun_201803211200 Daylight at 12:00 on 21st of March
sun_201803211400 Daylight at 14:00 on 21st of March
sun_201803211600 Daylight at 16:00 on 21st of March
sun_201803211800 Daylight at 18:00 on 21st of March

Summer Solstice

dimension description
sun_201806210600 Daylight at 06:00 on 21st of June
sun_201806210800 Daylight at 08:00 on 21st of June
sun_201806211000 Daylight at 10:00 on 21st of June
sun_201806211200 Daylight at 12:00 on 21st of June
sun_201806211400 Daylight at 14:00 on 21st of June
sun_201806211600 Daylight at 16:00 on 21st of June
sun_201806211800 Daylight at 18:00 on 21st of June
sun_201806212000 Daylight at 20:00 on 21st of June

Winter Solstice

dimension description
sun_201812211000 Daylight at 10:00 on 21st of December
sun_201812211200 Daylight at 12:00 on 21st of December
sun_201812211400 Daylight at 14:00 on 21st of December
sun_201812211600 Daylight at 16:00 on 21st of December

Noise / Window Noise

Noise level and the distribution of elements from an area help to understand how an object is exposed to the acoustics of this area. The acoustic simulation calculates the noise intensity on each individual hexagon from the analyzed object considering traffic and train noise datasets. Adjacent buildings are considered noise-blocking elements. The values are expressed in dBA (decibels).

Window Noise

The noise per window of a given area is aggregated via min and max. For instance, window_noise_train_day_max represents the maximum amount of noise received on any window of the area.

dimension description
window_noise_traffic_day The amount of noise received on the area’s windows from daytime car traffic
window_noise_traffic_night The amount of noise received on the area’s windows from night-time car traffic
window_noise_train_day The amount of noise received on the area’s windows from daytime train traffic
window_noise_train_night The amount of noise received on the area’s windows from night-time train traffic

Area-Wise Noise

The area-wise noise describes the amount of noise received from a noise source aggregated over the whole area in an unaggregated form. For instance, noise_traffic_night describes the dBA of noise received in the area from car traffic at night when propagating noise from all windows.

dimension description
noise_traffic_day The amount of noise received in the area from daytime car traffic
noise_traffic_night The amount of noise received in the area from night-time car traffic
noise_train_day The amount of noise received in the area from daytime train traffic
noise_train_night The amount of noise received in the area from night-time train traffic


Connectivity

Centrality simulations help to analyze a floor plan, whether it’s a shopping mall and you want to identify prominent areas in order to select the most prominent spot or it’s an interior design circulation path and you want to determine open floor plan areas. Centrality simulations are done using topological measures that score grid cells by their importance as a part of a grid cell network.

The distances and centralities are aggregated via minmaxmeanstdmedianp20, and p80. For instance, connectivity_balcony_distance_min describes the shortest distance to the next balcony from the point closest to the balcony in the area.

Distances

dimension description
connectivity_room_distance Distance to the next area of type ROOM
connectivity_living_dining_distance Distance to the next area of type LIVING_DINING
connectivity_bathroom_distance Distance to the next area of type BATHROOM
connectivity_kitchen_distance Distance to the next area of type KITCHEN
connectivity_balcony_distance Distance to the next area of type BALCONY
connectivity_loggia_distance Distance to the next area of type LOGGIA
connectivity_entrance_door_distance Distance to the next apartment exit

Centralities

dimension description
connectivity_eigen_centrality The Eigen-Centrality value
connectivity_betweenness_centrality The Betweenness-Centrality value
connectivity_closeness_centrality The Closeness-Centrality value

Location Properties

In addition to the apartment-related data, we also provide simulation data on the climatic, and infrastructural characteristics of the locations. The file locations.csv contains the simulation data aggregated on a per-building basis. Each row contains the identifier building_id corresponding to the building ids referenced in geometries.csv and simulations.csv.

Climate

The climate features represent 39 simple features based on the spatial climate analysis of Meteo Swiss as derived from MeteoSwiss.  Each column is formatted as climate_<category>_<period>. For instance, the column climate_tnorm_january  describes the monthly mean temperature in degrees Celsius (from the norm period of 1991-2020) at the location of the building. The aggregation methods vary per simulation category and are described in detail below.

Temperature Normals

dimension description
climate_tnorm_year The yearly mean temperature in degrees Celsius of the current norm period from 1991 to 2020 (TnormY9120)
climate_tnorm_january The monthly mean temperature in January in degrees Celsius of the current norm period from 1991 to 2020 (TnormM9120)
climate_tnorm_februry The monthly mean temperature in February in degrees Celsius of the current norm period from 1991 to 2020 (TnormM9120)
... ...
climate_tnorm_december The monthly mean temperature in December in degrees Celsius of the current norm period from 1991 to 2020 (TnormM9120)
climate_tminnorm_january The monthly minimum temperature in January in degrees Celsius of the current norm period from 1991 to 2020 (TminnormM9120)
...  
climate_tminnorm_december The monthly minimum temperature in December in degrees Celsius of the current norm period from 1991 to 2020 (TnormM9120)
climate_tmaxnorm_january The monthly maximum temperature in January in degrees Celcius of the current norm period from 1991 to 2020 (TnormM9120)
...  
climate_tmaxnorm_december The monthly maximum temperature in December in degrees Celcius of the current norm period from 1991 to 2020 (TnormM9120)

Sunshine Duration Normals

dimension description
climate_snorm_year The yearly mean relative sunshine duration in percent of the current norm period from 1991 to 2020 (SnormY9120). Relative sunshine duration (RSD) is the ratio between the effective sunshine duration and the duration maximally possible if no clouds were covering the sun. A period with sunshine is defined as a period when the direct solar irradiance exceeds 200 W/m²
climate_snorm_january The monthly mean relative sunshine duration for January in percent of the current norm period
climate_snorm_februry The monthly mean relative sunshine duration for February in percent of the current norm period
... ...
climate_snorm_december The monthly mean relative sunshine duration for December in percent of the current norm period

Precipitation Normals

dimension description
climate_rnorm_year The yearly mean precipitation in mm of the current norm period  (RnormY9120)
climate_rnorm_january The monthly mean precipitation for January in mm of the current norm period  (RnormM9120)
climate_rnorm_februry The monthly mean precipitation for February in mm of the current norm period  (RnormM9120)
... ...
climate_rnorm_december The monthly mean precipitation for December mm of the current norm period  (RnormM9120)

10-Minute Walkshed Infrastructure

Based on OpenStreetMap data and its tagging system we counted all 465 tags (key and value tuples as listed here: https://wiki.openstreetmap.org/wiki/Map_features) which can be reached within a 10-minute walk from the location of the building. Each column is formatted as walkshed_<poi_category>_<poi_type>. For instance, the column walkshed_shop_coffee  describes the number of coffee shops located within 10 minutes of walking from the building.

The following is an excerpt of support categories and their corresponding types.

  • shop: antique, art, ...
  • amenity: art, atm, ...
  • tourism: alpine, attraction, ...
  • leisure: amusement, beach, ...
  • healthcare: clinic, dentist, ...
  • historic: archaeological, battlefield, ...
  • ariaelway: station

Location Ratings

The location ratings, provided by Fahrländer Partner AG, give insights into the living situation at locations in which the buildings are situated. The file location_ratings.csv provides the following information:

dimension description
location_rating_MIKRAT_W Living situation - Overall (1.0=worst, 5.0=best)
location_rating_IMAGE_W Living situation - Image (1.0=worst, 5.0=best)
location_rating_DL_W Living situation - Service Quality (1.0=worst, 5.0=best)
location_rating_FZ_W

Living situation - Leisure Quality (1.0=worst, 5.0=best)

location_rating_NASE_W_DOM The most dominant segment of demand:

1 Rural-traditional
2 Modern worker
3 Transitional-alternative
4 Traditional middle class
5 Liberal middle class
6 Established alternative
7 Upper middle class
8 Professional elite
9 Urban elite
10 Unknown

More Information
location_rating_FGFRQZ

The mean number of pedestrians per hour throughout a day between 7 am and 8 pm of an average working day.

1 <50
2 50-100
3 100-200
4 200-500
5 >500

Files

swiss-dwellings-v3.0.0.zip

Files (931.9 MB)

Name Size Download all
md5:3b7915ecd5bf8e492a3e78c598b74123
931.9 MB Preview Download

Additional details

Related works

Is compiled by
Software: https://github.com/Archilyse/Archilyse (URL)
Is referenced by
Other: https://archilyse.github.io/ (URL)