Need to know for AIRBNB hosts
How Airbnb data help us
Millions of people are using Airbnb as hosts or guests from all around the world. There are more than 6 million listings in more than 100000 cities
With this post, I will provide some data-driven insights to the current and prospective hosts to help them improve their experiences by answering three questions
- How prices change according to time and neighborhood?
- Which factors help us to predict the price?
- Which factors help us to predict customer satisfaction?
To show it in an example, I used Airbnb Seattle 2017 data. There are 3818 listings, and data involve information related to place, host, and neighborhood as well as the date, price, and review scores. You can access the data sets from here.
Part 1. Understanding the Trends
First things first: To understand data, we need to describe it. Two things seem important to me: time and neighborhood. Let’s take a look at both…
The graph below shows the mean prices from January to December 2017. This graph tells us the prices reach their highest rate in summer, and they are the lowest in February. This data is aggregated over the date, but it still shows us a trend. So you may think of updating your prices according to seasons.
Our second graph shows the mean prices within each neighborhood. As shown in the picture, The most expensive area in Seattle is Magnolia, and the least expensive one is Delridge. The mean price difference between these two is around 100$ per night, which seems quite a difference.
Part 2. Predicting the Price
What else would affect the price? The chart below shows the prediction coefficients in the regression analysis. Simply, regression analysis tells us how changes in some variables affect our target variable (in our case, it is the price). So, this chart shows us how changes in the variables on the vertical axis increase or decrease the price. If you want to learn more about how regression works, this post may help you.
Note that the numbers in the graph are normalized, so they are not direct magnitude in price changes. This being said, I have four conclusions from this chart:
1. The number of bathrooms and bedrooms are the two most effective predictors of the price.
2. Houses are less expensive than apartments and other types of properties.
3. Interestingly, the places whose hosts put a profile picture are cheaper than the ones who don’t.
4. Being a super host increases the price.
The first finding is straightforward that as the increase in bedrooms, bathrooms, and accommodations increases the price. The third one is unexpected.
We may not change the nature of our place, but you may still have something to do to increase your price: Becoming a super host!!!
One condition to become a super host is having a rating of 4.8 (Airbnb). If you want to learn more about increasing your guests’ satisfaction, take a look at the next part.
Part 3. Predicting the Satisfaction
Finally, which factors affect our guests’ experiences? Our last chart is similar to the third one, but this one predicts satisfaction.
Just a reminder, the numbers in the chart are normalized, so they do not show the actual magnitudes. I will give you three takeaways from this chart:
- Your guests’ experience with your way of communication and the cleanliness of your place are the two most important factors that determine their general satisfaction.
- If your identity is verified and you use a flexible cancellation policy, your guests are more likely to be satisfied
- Interestingly, the price has almost nothing to do with the satisfaction
Conclusion
In this article, we took a look at Airbnb 2017 Seattle data to get a deeper understanding of the business. We described the data first. Then we used linear regression to understand what factors help us to predict the price and customer satisfaction.
Note that the findings here are observational, not the result of a formal study.
To see more about this analysis, you can take a look at my GitHub repository here.
