housing market activity over space and time

Explore the real estate market activity in the Canton of Zurich over space and time

The events on the real estate market are followed meticulously by many. No wonder that the property transfer statistics are among the top sought-after statistical data sets in the Canton of Zurich. At the TWIST-hackdays, for the first time, an open test-data set that includes all real estate transactions in the period 2010-2014 is being made available to the public.

You get a first-time opportunity to work with a high-resolution spatial dataset of real estate trades. The dataset consists of single trade entries with specifications about the construction year, the number of rooms as well as the living and property areas. These information are given in categories. Furthermore, a spatial component is given in form of a hexbin ID which allows you to join the data to a hexbin grid with a 1 square kilometer resolution.

Explore real estate market activity spatially and over time

The data set allows to follow the real estate market over the years 2010 to 2014. It can thus be used to analyze the activity on the housing market over space and time. Where has the market been expecially liquid - where could the most apartment or single-family homes trades be observed?

As we’ve also prepared a data set containing information on the stock of housing over the same hexagon-grid, the trading-data can be set into relation with the number of potentially tradable objects. What patterns come to light when the number of trades are set into relation with the number of existing housing units within the same area? It might as well be intriguing to find patterns and to provide spatial explanations by recurring to other open data sets.

Help us understand whether data sets in this new form are useful and save

We are eager to get feedback on whether such detailed single-record data meets user needs. We’ve alleviated the sensitivness of the highly detailed data. The most important step consisted in replacing the original geocoordinates with approximate information on where the object is located (hexagonal-cells of 1 square km). Further we’ve conducted tests for uniquenes and categorized variables to further lower. Help us improve, test the data for flaws and tell us how we could do better.

Data download and join with the hexbin information

The dataset consists of single data entries with location-information in aggregated form. It can thus easily be processed with R or other programming languages. By aggregating and matching the data to the hexbin-shapefile the data can also be analyzed with GIS-software. The following code chunks help you with the download of the data and demonstrate how the data can be joined to the hexbins to get the spatial information in R.

The Hexbins need to be downloaded and unziped seperately. You can read in the shapefile with the read_sf()-function from the sf-package.

You can find the zip-file under the following link: https://www.web.statistik.zh.ch/twist/Hexbins_1sqkm.zip

You can find a data-description file under the following link: https://www.web.statistik.zh.ch/twist/Data_description.pdf

example code for R

# load packages
# load the data

single_familiy_home_transactions <- read.csv("https://www.web.statistik.zh.ch/twist/total_dataset_efh_en.csv")

apartment_transactions <- read.csv("https://www.web.statistik.zh.ch/twist/total_dataset_STW_en.csv")

total_house_stock <- read.csv("https://www.web.statistik.zh.ch/twist/GWR_Datensatz/total_dataset_GWR.csv")

hexbin <- read_sf(paste0("path","Hexbins.shp"))
# join the transaction data to the hexbins

EFH_hexbin_info <- hexbins_reduced %>% left_join(single_familiy_home_transactions %>% dplyr::filter(year==2010) , by = "GRID_ID") %>% st_as_sf(crs=2056)

STW_hexbin_info <- hexbins_reduced %>% left_join(apartment_transactions %>% dplyr::filter(year==2010), by = "GRID_ID") %>% st_as_sf(crs=2056)

Let’s take a first glimpse at the data!

# aggregation of the single-family homes data

EFH_Hexbins <- EFH_hexbin_info  %>% dplyr::group_by(GRID_ID) %>% summarize(Number_EFH_transactions = sum(!is.na(recordid)))

# plot the number of single-family homes transactions for the year 2010 data

p <- ggplot() + geom_sf(data = EFH_Hexbins, aes(fill=Number_EFH_transactions) ) + ggtitle("Number of single-family-homes transactions of the year 2010") + coord_sf(datum = NA)