Finding an apartment in Lausanne with a rent that fits a student budget is a time-consuming and tedious task. If you have just arrived in Lausanne, you don’t know where to look. Is it cheaper on the hill or near the university? You don’t know all the parameters that come into play and while running from one flat viewing to another you start to wonder: who owns all this real estate you would like to live in?
We want to bring some light into the opaque world of real estate. By leveraging openly accessible data our goal is to understand how rent prices are influenced by the type of the owner, the geographical situation, or other factors. This article should serve as a head start for people searching affordable rents in Lausanne.
In particular you will be enlightened by the following real estate insights and you will understand how these were obtained. Even if Lausanne is not New York we needed to think of it in smaller units. Therefore, our results divide the city into its neighbourhoods – in french quartiers.
The owner with the most parcels in Lausanne is the municipality.
The quartiers Montriond and Ouchy are clearly more expensive than the rest.
The city centre is mostly owned by corporations, Chailly is mostly owned by privates.
There is no direct relation between owner type and prices. The market seems to adjust prices uniformly over the ownership types.
The price per square meter of a flat is strongly influenced by the distance to the lac Léman and by the surface of the flat. Smaller flats cost more per square meter than large flats.
Our first dataset consists of the geographical, cadastral and address data behind map.lausanne.ch. It features the owner of each of the almost 8000 plots or parcels (of land) in Lausanne. There are about 40001 entities possessing real estate and they are as diverse as you would imagine them, ranging from the municipality, to private people, to even multinational companies like Crédit Suisse, Phillip Morris International or the pension fund of Swatch.
The owner with the most parcels is unsurprisingly the city of Lausanne. With 1265 parcels it owns 12% of all parcels. This is ten times more than the next two owners which are the pension funds of the city and of the canton Vaud. Because most owners only have a small amount of parcels we will group them into 7 types:
public institutions: the city, the swiss railways etc.
pension or similar funds: investment foundations, the city’s pension fund, etc.
corporations: listed public companies like Swiss Life S.A., Régie Chamot & Cie S.A. etc.
cooperatives: registered cooperative companies like Migros, la Mobilière etc.
foundations and associations: for example the olympic foundation for cultural heritage.
PPE: single flats in a building owned by different private people – in french proriété par étage.
individual privates: private citizens owning an entire building.
If we look at the data as a map a very noisy mosaic shows up.
If you just squinted while looking at the map, we have the same intuition. The mosaic is too chaotic to say anything. Therefore, we try to smoothen the picture – digitally. For each parcel we drew an imaginary circle through its neighbouring parcels and looked at their ownership type. The cell was then reassigned to the type which covered the most of the circle’s surface. This can be seen as a weighted k-nearest-neighbours algorithm with variable k. (Click on the legend of the above map to see the second layer.)
With this cleaner picture, some patterns emerge. The eastern quartier Chailly is dominantly owned by privates. Big parks, the lakeside and the rail lines are of course possessed by public institutions and the centre of the city has the highest density of corporations.
The fact that the two maps are different shows that there is a lot of diversity in some quartiers’ ownership patterns. In order to see which quartiers are the most diversely owned, we computed another map that measures the diversity with the Shannon entropy of the owners in a circle around each parcel.
Unsurprisingly, the large parks, the airport and the lakeside which all belong to the city have low entropy in their ownership patterns. That means there is a local monopoly of owners. While this was expected for the city’s properties it is rather a discovery for the Flon in the centre of the city. This area is red as well because a single corporation (LO Immeubles S.A.) owns all of the properties in the Flon!
If you visit a real estate portal you see something like this. You don’t get a global view of the area and its prices. To overcome this, we collected the listings from the three most important swiss platforms (Anibis, Homegate and Tutti). After removing duplicates and fake offers this gave us 469 offers with prices in CHF/m2. By combining them with our geographical data from before, we can present them in the map below.
These points don’t really help if one needs information for a property between two points. Therefore, you find the second map that features the median rent price for each quartier. (Click on the second layer of the map.)
We were still not satisfied by this second map, because it averages out all fine-grained information. The k-nearest-neighbours algorithm seemed like the perfect match for the problem. (And not only because of its name.) By looking at the nearest neighbours of a parcel we predict its rent price. This allows us to get a smooth heat-map of rents in Lausanne which is displayed as the third layer below:
There seem to be two hot spots for high rents: the city centre and even more significantly the two quartiers near the lake and the port Montriond and Ouchy. For our student readers, the affordable rents are more on the hill, for example near the airport.
Having seen the differences in ownership across the city, we ask ourselves whether the ownership type also influences the price of accommodation. To give a statistically correct answer, we evaluated a simple linear model with linear regression. The model tries to find significantly different mean rent prices depending on the ownership type of a parcel. This model however does not give a meaningful result. It seems like our hypothesis that rents are different if the property is for example owned by a corporation has no factual basis, and that the small differences in the mean rents of each ownership type just arise from random noise. After all, we only have a small number of offers compared to the number of parcels.
But if the owner doesn’t influence the price, what does? Why are prices different? You can get an intuition for this by looking at the smoothed rent map. Clearly, the parts near the lake are more red. Our next linear regression therefore analyses the dependence of the price on the distance to the port of Lausanne, Ouchy. As you can observe in the plot to the left, there is a visible correlation. Our model can confirm with a high significance that the more you go towards the country-side, the cheaper flats get.
There is also a second factor that influences rent prices per square meter. This aspect is less obvious: the surface of the flat inversely correlates with the price per square meter. From the plot one can see that small flats have a very high price per square meter.
Even if we did not exactly find what we set out to find, we learned quite a bit about rent prices and quartiers. Our first hypothesis that rent prices are influenced by the owner was rejected. We see two reasons for this: First, we only had sparse data about rents. The circa 500 offers we collected only cover a small percentage of rental units in Lausanne.
The second and more important reason is the market. By the law of supply and demand prices will always balance one another. Put in other terms, no owner will offer their property at a higher price than the neighbours because this would be an economical disadvantage. Similarly, if all neighbours offer their real estate at a high price the owner will do the same. This demonstrates the importance of the quartiers or in general the geographical situation that we showed was a significant factor for the price.
As students at EPFL we know how hard it is to find accommodation in Lausanne. We hope that this analysis helped you understand where to look for a flat and what parameters determine the prices across the city. This projects also shows the potential of open data for providing transparency and insight into otherwise opaque systems. We can only encourage to do the same for other things that interest you!2