What determine the Airbnb Price? A look at Boston and Seattle Data
Are you ever wondering what features of Airbnb rental homes drive up the price?
The purpose of this article is to provide some insights on the Airbnb pricing in Seattle and Boston. In particular, three questions are investigated:
- Are there significant difference between Boston and Seattle Airbnb data?
- What features are most dominant to determine the rental price in both markets?
- What amenities are import to determine the rental price in both markets?
Are there significant difference between Boston and Seattle Airbnb data?
Boston Airbnb data has 3585 entires in the listing. The average price of Boston Airbnb is $174/night, with standard deviation of $148. The 50% of Boston listing are equal or below $150/night, and 75% of listings are below $220/night. 99% of the listing lie under $350/night. The most expensive listing in Boston is $4000/night and is located at Commonwealth Avenue in Feenway park neighborhood.
While Seattle Airbnb listing data has 3818 listings. The average price of Seattle Airbnb is $128/night, with standard deviation of $90. The 50% of Seattle listing are equal or below $100/night, and 75% of listings are below $150/night. 99% of the listing lie under $275/night. The most expensive listing in Seattle is $1000/night and is located at 7th Avenue West in Queen Anne neighborhood.
As shown the in the figure, both cities have the listing price concentrated between $50–$300. Compared to Boston market, Seattle prices follow more a normal distribution.
Hence the data suggests that it costs you more to travel to Boston than to Seattle in terms of accomodation.
What features are most dominant to determine the rental price in both markets?
In order to determine the features that contribute to the Airbnb pricing, I used an Ensemble supervised learning method, Random Forest Regressor to fit the historical data.
The purpose is to use the feature_importances_ attribute to give an indication of feature to price relationship. To focus on majority of the listings, I decided to keep only the price within the 95 percentile, which eliminates the anomaly data.
I listed the 20 top features in both markets that have the most weights in the feature. It makes sense that the size of the house/apt has the most impact on the rental price in both markets. The top two features in both markets, Roomtype_entire_home/apt and bedrooms are indicators of the size. The next influential features in both markets are the location and zip code, as we all know that the location is a primary factors in real estate market. The data confirms this.
What amenities are import to determine the rental price in both markets?
I looked at the top amenities that contribute to the rental prices. First observation is those amenities are not highly impacting the rental price as the most impact of amenities is about 0.8% to contribute in the pricing. None of them cracked the top 20 price relevant features.
However, it does show relatively what amenities in determining the price. The most important amenity in both markt is the elevator in the building. It makes sense as we are looking two major urban areas where there exist many high rise buildings. Then amenities like breakfast, gym, start to show on the list.
It should be mentioned that Random Forest Regressor model trained with Boston Market data yield a R2 testing socre about 60%, while it only gives about 30% R2 testing score on Seattle data. It should be further investigated.
Conclusion
In this article, we dive into the most recent Airbnb Boston and Seattle dataset and found many interesting insights.
I preprocessed the data, columns with many missing values are dropped and NA values are taken care of individually.
I trained a Random Forest model to predict the rental price. It should be noted that the model gives an Okay R2 testing score for Boston listings but a poor score for Seattle (0.3). This should be investigated or tuned more.
Inportant features and amenities to decide the rental price are listed.