AirBnb Price Prediction
The rise of Airbnb has fundamentally reshaped the Boston accommodation landscape. Initially, it emerged as a solution for residents facing high rents, offering a way to unlock the value of their unused space through short-term rentals. However, as Airbnb's popularity soared, it transformed into a dynamic marketplace catering to a diverse range of travelers seeking unique experiences beyond traditional hotels.
This project delves deeper into the intricate data ecosystem of Boston's Airbnb listings. By employing advanced machine learning and data mining techniques, we aim to illuminate key factors that influence listing prices and guest preferences.
This multifaceted analysis will not only equip Airbnb with valuable insights to optimize its platform for both hosts and guests, but also empower potential Bostonian hosts with data-driven knowledge to make informed decisions about their rental strategies. Ultimately, this project strives to bridge the information gap and foster a mutually beneficial environment within the Boston Airbnb market.
During data preprocessing, we evaluated the usefulness of both boolean and categorical features. The goal was to identify features with a limited number of categories, potentially indicating insufficient data for meaningful analysis.
Several columns, including 'calculated host listings count shared rooms', 'review scores location', and 'calendar', exhibited this characteristic. These features with limited categories were deemed less informative and subsequently removed from the dataset.
The figure shows that jurisdiction_names, liscense, neighbourhood_group_cleansed and has_availability almost all null values and hence it won't be of any use to us for the analysis.
The features like bedroom, beds, and accomodates are highly correlated features with the target variable 'Price'
The figure shows us the correlation plot of all the features in the dataset. All the highly correlated features were removed so that it won't affect out machine learning models.
The given interactive map shows the airbnb listings in the greater Boston area and as we can see, most of the listings are located in the Central Boston and near Back Bay
By the neighbourhoods, Jamaica Plain and South end followed by back bay have highest listings
As seen in the figure, the most expensive neighbourhood in the Boston is the South Boston waterfront, this must be because of the houses which are facing the ocean and the prices are high.
It is followed by Bay village and leather district.
The hosts joining the airbnb had peaked during late late 2016 and late 2020. This might be because a lot of people must have travelled after the Covid Lockdown restrictions were cool down.
As shown in the figure, the most important features in deciding the price of a home are the number of bathrooms, is the entire home is available or not, number of people it can accommodate and whether it has an access to gym or not. They have the feature weights as - 0.09,0.20,0.08 and 0.03 respectively.
The other important features are parking space availability and the number of nights the Airbnb property is available.
The most important features in deciding the price of a home are the number of bathrooms, is the entire home is available or not, number of people it can accommodate and whether it has an access to gym or not. They have the feature weights as - 0.09,0.20,0.08 and 0.03 respectively.
Using the XGBoost model, we can predict the price with an RMSE of 96.488. Alog with this, other important features are parking space availability, minimum number of nights the property is available, and whether the host provides response or not.
We can conclude that most of the listings are around central Boston and the most expensive neighborhood is the South Boston waterfront area with an average price more than $300 per night.
So, by using various machine learning models we can help the hosts and Airbnb to predict the best price range for any property in the Boston area.