Analyzing Berlin’s Neighbourhoods with respect to Host Locality using Airbnb Data.
Do you also live in the coolest neighbourhood of your town? As do I!
Speaking with people it seems like everyone is living in the best part of the town. Usually you value one part of the town more than the others and are willing to pay more to live there. But does this effect also go the other way around?
Do hosts living in the same neighbourhood as their rentals are located charge higher prices than average?
To answer this question we are going to use the airbnb dataset and analyze it with regard to neighbourhoods, prices and room types.
What is the average price of Listings per Neighbourhood?
First of all we have to determine what the average price per neighbourhood is. Therefore the data is grouped into twelve neighbourhoods. In order to estimate validity of the results, let us start with looking at the sample size of each neighbourhood:
Tempelhof - Schöneberg 1380
Treptow - Köpenick 649
Steglitz - Zehlendorf 438
Marzahn - Hellersdorf 144
We can clearly see that there is a great variaty between the neighbourhoods. Though with sample sizes above 100 for all of them we should be able to trust the results when using the whole data.
The most expensive neighbourhoods are Mitte, Marzahn-Hellersdorf and Spandau. While Mitte has the largest amounts of listings, the other two have the lowest number of listings.
The least expensive neighbourhoods are Reinickendorf, Neukölln and Lichtenberg.
Do local hosts charge more for their Listings than average?
To answer this question, we have to match host neighbourhoods and listing neighbourhoods. Since both are grouped differently, we start by comparing an amount of 18% local hosts to the average and six of the neighbourhoods:
- Only Mitte is showing higher prices for local hosts.
- Neukölln is very balanced while the rest is mostly showing slightly higher prices for the average of all listings.
- In Spandau local hosts only charge half of the average price of all listings.
- But since Spandau had the smallest sample size in the whole set, we should take a look at the amount of local host listings
The sample of Spandau in the local host mask is very small, therefore unlikely meaningful.
We see that many of the neighbourhoods of our original listings are not matched. Next, we will try to improve the number of neighbourhoods matched.
Improving the amount of matched neighbourhoods:
Looking at the matched neighbourhoods names, we can clearly see that the hyphen-separated ones are missing.
Therefore we search in substrings for the neighbourhood names with the success of now matching 42% of the neighbourhoods. Resulting in the following counts:
Tempelhof - Schöneberg 864
Steglitz - Zehlendorf 156
Marzahn - Hellersdorf 49
Treptow - Köpenick 42
More details regarding the implementation of this task can be found in the GitHub linked at the end of the article.
Can we get a clearer picture of local Host Prices looking at all twelve Neighbourhoods?
Only the hyphen connected neighbourhood-names are added to the plot, namely:
- Friedrichshain-Kreuzberg and Tempelhof-Schöneberg which show very leveled prices
- Charlottenburg-Wilm. and Steglitz -Zehlendorf while also being pretty close together is slightly in favor of all hosts
- Marzahn-Hellersdorf, Treptow-Köpenick have a pretty low sample size below 50. Nonetheless they are favoring local hosts
Unforunately, we still see no clear signs either that local hosts charge higher or lower prices in all neighbourhoods.
Questions that come to my mind
- do local hosts offer more private rooms than other hosts?
- do private rooms tend to be cheaper?
Local Hosts do not clearly offer an above average Ratio of Private Rooms
Maybe local hosts tend to offer parts of their own flat more often. This could be seen by looking at the room type of the listing. Differences in this category could lead to the statement that local hosts charge nearly equal prices while offering a different room type.
As before, the questions can’t be clearly answered for all neighbourhoods.
It seems that local hosts are slightly offerering an above average number of private rooms, though the difference is not really significant in most neighbourhoods.
Interestingly, the bars of the plot are pretty much reversed from the plots before. E.g. Mitte has an below average count of of private rooms for local hosts, while they had the highest prices for local hosts at the same time. While Spandau was the complete opposite. Other Neighbourhoods have the same behaviour.
This strengthen the assumption that private rooms tend to be cheaper. We still cannot see any signs that local hosts charge more than the average host.
Private Rooms indeed are cheaper than Entire Homes/Apts
Classification of the room type in the data:
Entire home/apt 11036
Private room 8868
Shared room 279
Hotel room 236
Since the vast majority of listings is either categorized as Private room or Entire home/apt we will only compare these two in the following.
Unsurprisingly, private rooms are indeed cheaper in all of the present neighbourhoods.
In this article we took a look at airbnb data of Berlin and analyzed whether local hosts charge higher prices for their listings.
We found out that at least 42% of the hosts live in the same neighbourhood as their listings in Berlin are located.
Those hosts do not exclusively charge an above average price for their neighbourhood but the results are pretty mixed.
Even taking into account the factor of room type we were not able to find any indications that they charge higher prices.
The one thing we were able to determine is that private rooms tend to be cheaper than Entire homes/apts.
Would you still charge higher prices because your neighbourhood really is the place to be?
The underlying code and analysis can be found in my GitHub available here.