How Do YOU Set Your Airbnb Price?
A data driven approach using Airbnb listings data for Vancouver from 2018.
Introduction
The price you charge for your listing is completely up to you. One common way is to search for similar listings in your city to get an idea of market price. Although this method has shown to work fine since emergence of Airbnb and other comparable services I aim at finding a scientific approach to set the price of Airbnb listings.
There are many factors that can affect the price : location, amenities, the host reveiw rating, being a superhost, number of beds, bathrooms, and rooms, square feet, to name a few.
There are many questions that can be asked when talking about Airbnb listings and their price:
You may see many amenities in a listing, but how important are those?
How much they can affect the price?
How listings are distributed in the city and in which areas listings are distributed densely?
You might be able to answer these questions if you have lived in the city. In fact, I thought I could answer all these questions without looking at the data! But what does the data suggest?
Therefore, I employed data from Vancouver Airbnb listings from 2018, to take a closer look at these questions.
The data covers 6437 listings from 4860 hosts with 96 features for each listing in the data set.
Part I:Is there a relationship between price of a listing and its review score?
Below you can see distrbution of listings in the city. The distribution is denser in downtown area. Outside of downtown area, Kitsilano beach has the highest density of listings followed by East side. South Vancouver has the lowest density of Airbnb listings.
Most of listings are in the range of $0.0 to $1000. There are few listings with an extreme price, higher than $700, which make only 1% of the data. I decided to take the data whose price are less than $700 for better model predictions and consistency.
Looking at Figure 2, It seems that a high price does not lead to a high review score necessarily. There are many listings with a price in the range $100 — $300 and with a very high review score. There are some expensive listings with a low review rating as well.
60.2% of the hosts have a review score greater than 90% while not listed as a super host. It seems being a super host is not necessary for owning a high review score. As we can see from the scatter plot, there are many hosts which are not classified as a super host according to Airbnb definitions, but yet have a high review rating. A little google search reveals the requirements for being a super host:
Hosted at least 10 trips
Maintained a 90% response rate or higher
Received a 5-star review at least 80% of the time you’ve been reviewed, as long as at least half of the guests who stayed with you left a review
Completed each of your confirmed reservations without cancelling
It seems that being a super host is too difficult! Maybe Airbnb can modify its requirements for being a super host!
Part II: What are the most important factors in determining the price of Airbnb homes?
To answer this question, I performed a data cleaning proccess on the listings data set by imputing missing values, parsing text features, normalizing the data. The clean data was fed to three Machine Learning models : Linear model, Decision Tree model, and Random Forest model.
The root-mean-square error (RMSE) was calculated for each of these models. Initially, the Decision tree gave the lowest RMSE. However, performing a 10-fold cross validation revealed that the Random Forest generates the best results. The selected model was fine tuned to obtain the optimum parameters.
As you can see from Figure 3. number of beds, bathrooms, and bedrooms are the most important features that determine the price. This is not surprising! we could expect that the price of an Airbnb can be affected significantly by the these features.
There are other factors which their effects on the price was predictable such as: accommodates, private room, guest included, entire home/apartment. Surprisingly, availability in a year and total host listing are among the important features.
Among all the locations, we only have downtown in the top 20 features which makes sense. Normally, prices are higher in the downtown area of Vancouver.
We should note that the feature importance only tells us which features are important in prediction of an Airbnb price. It does not tell us anything about how variation of a feature varies the price. To study the sensitivity of price to each feature, one need to generate partial dependence plots.
Part III: What amenities of an Airbnb affect the price significantly?
Every day people add more and more amenities to their listing and hope that those extra amenities will improve their listing. There were 123 amenities in total. Figure 4 presents the important amenities that can affect the price.
Renting a Family/kid friendly Airbnb is an important concern for most of families. Therefore, it makes sense that this feature affects the price significantly compared to the listings which do not have this.
Indore fireplace is an extra cost for the host of an Airbnb place. So, having this feature normally increases the price. The same logic will apply to the rest of the features in the top 10 important amenities except TV.
It is surprising that TV can affect the price. It makes sense that Cable TV would change the price but normally most of Airbnb listings have TV and I would not count that as a special amenity.
To support my point, we can see that Wi-Fi is not among the top 10 amenities. It is actually 68 in 123 available amenities. That’s because almost all homes have Wi-Fi nowadays! It was expected that Wi-Fi would not affect the price significantly.
Conclusion
In this article, we took a look at important features that determine the price of Airbnb listings, how the listings are distributed in the city, and we explored the relation between the price and the rating score of Airbnb listings.
1. We gathered the Vancouver Airbnb listing data, which showed that the listings are denser in downtown area, with a price range up to $700.
2. We then looked at the relation between the price and the review score. Their scatter plot showed that high prices does not certainly lead to a high review score rating. In fact, there are many more affordable listings with a very good review score.
3. Finally, we looked at the important amenities that determine the price. The results suggested that number of beds, bedrooms, and bathrooms are the most determining features in predicting the price.
The findings here are observational, not the result of a formal study. So, the real question remains:
How will YOU set your Airbnb price?
To see more about this analysis, see the link to my Github available here.
