PATRIZIA extends Big Data intelligent solution to the Netherlands
PATRIZIA is expanding its Big Data analytics’ solution to the Dutch residential property market, allowing intelligent processes to gather and analyse local real estate data.
Deep insights into Dutch residential properties
At the end of the day, all we need is an address, declares Marcelo Cajias, Associate Director of Research at PATRIZIA. It’s a simple piece of information but one that can provide a host of valuable insights about the local residential market. At least if it’s run through as input into PATRIZIA’s Big Data analytics tool that is.
Now, following five years of success in Germany, PATRIZIA is introducing the tool to the Dutch residential property market.
“The location is the only input we need,” explains Cajias. “Our colleagues in the Netherlands send us an address, with a street, a street number, the city and the zip code. Our analytics’ machine then downloads automatically the data for this particular location. We program it all into the cloud and receive the results as a dynamic webpage for every single location.”
Using cloud-based systems, PATRIZIA’s Big Data analytics tool then compresses and analyses unstructured data from several multiple listing systems into one big database. This quickly delivers in-depth market insights. These can be used as the basis for sound investment decisions.
How it all works
So how precisely does the process work? The first step is to log into the database, select, clean and analyse the data and then aggregate the results, explains Cajias. This process is repeated many times using multiple data sources.
Recently, PATRIZIA has been adding geodata from sources including Eurostat, Oxford Economics and Statistics Netherlands (CBS), a Dutch governmental institution that gathers statistical information about the Netherlands. PATRIZIA then feeds this data with Google Maps and OpenStreetMap via APIs. PATRIZIA also employs localized socio-economic and socio-demographic information, which helps delivers information on aspects such as prices and rents.
Once the cloud system has gathered the information, PATRIZIA prepares and digitalises the data to allow colleagues and decision-makers to gain an overview of a particular market.
In the time it takes to drink a coffee, we can generate a wealth of data and insights on an individual Dutch residential property.
Dr. Marcelo Cajias
The next step is to understand prices and rents for the location in question. Machine-learning methods and advanced econometric tools enable PATRIZIA to break down prices into single attributes and assess individual factors determining prices, such as location, the number of rooms or the size of the property.
For the Dutch market, it’s especially important to differentiate between houses for rent or to buy and apartments to rent or to buy, adds Cajias. Evaluating the composition of prices then leads to intelligent insights on housing demand, housing supply, price indices and rent indices. What’s more, the tool can accurately estimate gross initial yields (the annual rent divided by the price) for single assets across the Dutch residential market.
And what’s most amazing is that all of this takes just four minutes and less than an hour for a portfolio.
Reporting on the details
Cajias demonstrates the speed and depth of the results by compiling a report about a property in northern Amsterdam. “We look at population growth for Amsterdam over time, population distribution by ages and the population forecasts, housing supply and official indices of house prices to gain an overview of the market.”
After just over four minutes the tool produces a dynamic web report comprised of several different layers: general information, information about the property’s location, local house prices, local house rents, apartment prices and rents, value drivers and the initial yields.
Possibly the most important parameter is the average house price and rent per square metre in a radius of 300 metres, 700 metres, 1000 and 2000 metres, which clearly demonstrates the current market.
In addition, the report additionally shows how prices are distributed according to the construction year, looking at newly built assets and older houses’ prices.
It’s interesting to see which variables most influence the prices in the surrounding area of the Amsterdam example. The main variable is the construction year – if a property is constructed prior to 1931 then the prices are higher as older buildings can mainly be found in the city centre. The second most important variable is whether or not a property includes a balcony.
The final decisive block of information derived is the gross initial yield, which mirrors what is happening in the market and changes from location to location. “We’ve learned from our experience with big data in Germany that machine learning methods work very well in understanding how initial yield vary from location to location and from segment to segment. This is why we track this indicator closely.”
PATRIZIA plans to soon add local information, such as where the next train station or supermarket is, to expand the tool’s capability and range.
“As the machine learns every day, the accuracy of the tool increases continuously,” says Cajias. “We believe that the information we provide an overview of the market and is influential in colleagues and investors’ decision-making process. It’s a dynamic view of the market, an overview of what is happening there.”