Airbnb Berlin: Pricing Factors & Effects on Local Market (09/2022–06/2023)
Challenge
Background: Berlin has had a tumultuous history with Airbnb, initially banning short-term rentals. While the ban was lifted in 2018, restrictions on non-primary residence rentals persisted. A 2021 study, funded by the public, indicated that long-term Airbnb rentals had driven up local housing costs. Airbnb introduced new rules, effective March 2023, mandating registration numbers from district authorities.
Objective: Identify the key factors influencing Airbnb pricing and ratings in Berlin. Monitor the changes in Airbnb listings following the March 2023 license mandate. Track and analyze the price variations in local rental costs in comparison to Airbnb pricing.
Sourcing open data
Data wrangling & cleaning
Geospatial analysis
Dickey-Fuller Test
Autocorrelations
Regression analysis
Clustering
Time series data sourcing and analysis
Supervised and unsupervised machine learning
Python
Jupyter
Pandas
NumPy
Matplotlib
Seaborn
SciPy
Tableau
Excel
Data assembled from Insideairbnb.com
Skills
Tools
Data
Pricing & Ratings
Key Question: What factors impact price and ratings?
Steps Taken: For this I generated a correlation heatmap in Seaborn (right)
Conclusion:
From correlation, the only quantitative factors affecting price were accommodation capacity and related comfort factors (bed counts, bedrooms, bathrooms, etc.), and at weak levels (0.5 and lower)
Ratings are only correlated with each other
Pair plots, scatterplots, regression tests, and histograms confirmed these results. Results were similar between complete and outlier-controlled versions of the dataset.
Using categorical plots, I then found the only qualitative factor influencing price was the kind of property. Reviews did not have such a factor.
Conclusion:
Property type is the primary qualitative determinant of price. The market is dominated by :
Entire Homes/Apartments (62.7% of listings; $117.56 on average per night)
Private Rooms (35.7%; $59.39)
Using unsupervised machine learning, I found that price is better explained by four distinctive clusters:
Small-Time Locals dominate (79% of listings), offering lower prices ($63/night) but fewer licenses (34%).
Central Property Professionals (18% of listings) have higher prices ($200/night), licenses (80%), and proximity to the center (4.32km).
Luxury Private Professionals are the priciest ($574.61) with larger accommodation capacity (6).
Distant Property Owners charge $150 per night, do not live in Berlin (0%) but most often mention licenses (73%) and have large listing counts (605)
Conclusion:
Price is primarily driven by room type and professionalism, as shown by the clusters.
March 2023 License Mandate
Steps Taken: I monitored the changes by creating dashboards in Tableau.
Conclusion:
More than 30% of Small-Time Locals delisted during this time, while Luxury Property Professionals grew 10%
All clusters greatly increased their mentions of licenses, though this also significantly declined between March and June for Small-Time Localss, indicating those concerned about the license requirement may have already left.
Despite the STL exodus, average prices for the different clusters overall remained flat.
Airbnb vs. Local Rental Costs
Key Question: How does Airbnb price growth compare to local rent price increase and inflation?
Steps Taken: I sourced rental price and inflation information from public sources and built Tableau dashboards to display the differences.
Conclusion:
Overall, Airbnb's prices are growing faster than both local rental prices as well as inflation
There are exceptions to both counts, especially in Marzahn-Hellersdorf and Spandau where prices decreased
If higher Airbnb prices do raise local rental prices, that trend is likely to continue. However, statistically, the correlation between Airbnb and local rental price growth rates is weak at 0.26
