🤖 Artificial intelligence, to make decisions and predict price🤖

AI to predict prices, lower value per square meter, higher value per rented night, and higher occupancy rate.

With our repression model, we seek to take market data to be able to make decisions about the best place to carry out a short-term rental project. We will take into consideration:

The lowest land cost, and the largest built space The lowest value per m2 nights rented per year the price per night The monthly occupancy rate The annual occupancy rate the average tourism rate per year The annual occupancy rate with tourism destination statewide

Architecture for optimizing prices per night.

Occupancy rate architecture based on ratings

Initially, we planned to have two separate models to predict the listing price and demand. One of those models used k-Means to cluster all Airbnb listings within a 3 mile radius to find groups of similar nearby properties, and took the average occupancy rate to estimate expected monthly demand within each group.

Then, we had another separate k-regression model, to forecast the daily listing price based on listing attributes, comparing the best amenities (comfort, security, etc). Finally, we estimated the expected monthly earnings by taking the average daily listing price for each month and multiplying it by the predicted monthly demand (the predicted number of days a property is booked in a month).

However, after many attempts to improve the accuracy of this model, we discovered it was unreasonable to predict demand and listing price separately, mainly because these two quantities are highly correlated with each other.

For example, demand is likely to decrease if listing price increases, since customers can be expected to make smart decisions and switch their booking to a cheaper property with similar features.

Therefore, we decided to first model demand and then incorporate it into our pricing function. This meant we needed to build a demand function that captured competitive factors, listing attributes, and time fluctuations.

We modeled the competitive factor using the same approach we initially planned to apply for our monthly demand estimation.

Customers tend to choose their Airbnb with a specific location in mind, so all listings that are located very close (within a 3 mile radius) are more likely to compete with each other. Beyond this, the characteristics and quality of the listing should have an almost equal, and occasionally greater, impact in determining occupancy rates relative to competing properties. Based on these guidelines, we chose to use k-Means to cluster similar properties within a 3 mile radius distance (figures 2 and 3).

Formulating the “competition” group’s demand function

For each group, we have a dataset with variable X being the listing price and variable Y being the demand represented by the occupancy rate.

From this, we will fit a linear or polynomial regression model on this dataset to find the best fitted demand function. We then plug this demand function into the optimization model.

The goal is to maximize profits over a year, so the formula for calculating daily profits is the listing price that day multiplied by the demand function that day, represented as .

It's worth noting that at this point in the analysis, the demand in our dataset is the occupancy rate, which represents the probability that a property will be booked on a given day.

We'll mention details of how to calculate optimal annual profits in the Decision Model later.

After several attempts, we determined that the relationship between demand and price is not always inversely correlated in a simple linear way; outliers in particular don't fit our model, meaning extremely expensive listings appear to have different demand rules governing them, which is reasonable but not accounted for in our limited model.

We have 5 years of experience modeling data for the short term rental business, based on decision making.

Here you can access version 1 of our Jupyter book.

297KB

Preprocess_and_Modeling.ipynb