Did you know that you can get public Airbnb data through Inside Airbnb?
Meaning you can download free data from cities around the world to make data predictions on pricing based on market conditions. Also meaning, you don’t have to guess what to price your Airbnb at during the era of coronavirus.
According to various sources, Airbnb rentals are actually on the up and up since June and as hosts rev up their listings to get back on track, they’re probably wondering what the future holds for their property.
If you are having trouble finding renters or the perfect market price during this pandemic, you can actually get the answers from public data.
This use case is useful for Airbnb hosts, property managers, or anyone looking to predict prices based on current conditions.
We’re Going to Take Historical Data and Apply It to Current Market Conditions
Airbnb’s resource center suggests taking new approaches to hosting, such as:
- Expand your usual booking window.
- Rethink your marketing plan, and adjust your rates.
BUT, other than a few sentences, the article doesn’t say what to adjust your booking window or rates to.
We’re going to show you an overview of how to predict these things with machine learning and use data predictions to guide your decision making. We’re also going to take into account current pricing trends during the era of Coronavirus using publicly available datasets.
REMINDER: You definitely don’t need to have a background in machine learning to make these predictions. Obviously AI is a no-code ML platform that makes data predictions effortless for everyone.
About the Datasets We Will Use
All the datasets we will use, will be publicly available. We use publicly available datasets to show what you can do with data and inspire creativity. We don’t use any personally identifiable information (PII) and require datasets with a minimum of 1,000 rows and 5 columns to ensure max accuracy of the algorithms.
Read more about our data requisites here.
The dataset we will use includes summary information and metrics for listings in New York City, collected once a month from January 1, 2015 to July 7, 2020.
Expanding Your Usual Booking Window
The Airbnb resource center does a good job of providing current trends for their hosts. This article says group travel is trending and guests are currently looking for entire homes to stay in (as of June 4, 2020).
The first column we will predict will be “price.” This will give us a baseline knowledge of what the predicted price is.
For this prediction, we uploaded this dataset and started predicting. Here are the columns we used and the ones we didn’t use. The ones we didn’t use weren’t useful to this prediction. We’re trying to get an accurate prediction of the price in relation to location, reviews, and characteristics of the Airbnb listing.
Let’s start predicting.
We are instantly shown the top driving factors on the price of Airbnb listings.
The top 5 drivers by impact include:
- The amount of reviews per month the listing receives
- The availability of the Airbnb during the year
- The total number of reviews
- The neighborhood the listing is in
- The minimum number of nights stayed
According to the data, hotel rooms are more expensive than entire home/apts. The predicted price for an entire home/apt is $205.301. This takes into account the months the country was in quarantine.
Next, we can see the predicted price by neighborhood in NYC. This is particularly useful if you have multiple listings across a city or want to compare neighborhoods.
Not surprisingly, the most expensive neighborhood group is Manhattan.
And then there’s the minimum number of nights stayed.
According to the data, the sweet spot for guests to stay is in between 33 and 52 nights for the maximum price the guest is willing to pay. In other words, the guest is willing to pay the most for 43 nights in the Airbnb at $268.47/night.
Let’s predict the same thing, but with a pre-covid dataset. You can follow the same process as above and compare a dataset from December 2019.
Interestingly enough, the top driver on price pre-quarantine was the minimum number of nights stayed. According to the data, the predicted price for 40 nights is $234.40/night, $34.07 less than the dataset including the months in quarantine.
So, that means the advice from the Airbnb resource center advice is true. Guests are willing to pay the most for stays that are ~40 nights and willing to pay more for longer stays in 2020
But, it didn’t give concrete numbers like this. After 52 nights, the price drops back down. The areas in which the graph drops off are due to missing columns by the dataset. This is common in dataset collection. This could be due to no guest staying for ~62 nights, the data unable to be recorded, etc.
Getting a More Specific Price for Your Airbnb Listing
Taking this one step further, you can get a more specific price using our Personas tool.
You can literally fine-tune the prediction based on your listing to ALWAYS now the predicted price. You can choose the attributes to create different scenarios (or personas).
We filled in the minimum nights at 42 to see the predicted price range for an entire home/apt Airbnb in the Brooklyn Heights, Brooklyn as an example.
You can make as many personas as you want and save them for later to use.
Learn more about creating personas here.
About the Prediction
This prediction used a Random Forest Regressor algorithm. To learn more about the algorithm, read more here.
Keep in mind that this dataset was downloaded from Inside Airbnb and was not collected by Obviously AI. For a data prediction model to be accurate, the data needs to be structured well. We automate cleaning and preprocessing tasks to make the prediction more accurate and offer a call with a dedicated data scientist to improve your predictions.
You Should Now Have a Better Idea of How to Price Your Airbnb During Covid
There’s two amazing things about predicting your Airbnb price:
- You can get the datasets for free
- You can use Obviously AI for free
Free is a great step in machine learning democratization and allowing non-technical people to make these kinds of predictions.
To begin making data predictions with Airbnb data, login here.