Airbnb Price Prediction Using Machine Learning and Sentiment Analysis

Pricing a property and evaluating the proposed price for a property are challenges that, respectively, owners and customers of Airbnb rentals face on a daily basis. This paper aims to create a model for predicting the price of an Airbnb listing using property specifications, owner information, and customer reviews for the listing. Owners and customers can use the resulting model to estimate the expected value of an Airbnb listing. Linear regression, tree-based models, K-means Clustering, Support Vector Regression (SVR), and neural networks are trained and tuned on a dataset of Airbnb listings from New York city, and the resulting models are compared in terms of Mean Squared Error, Mean Absolute Error, and R-2 score. Sentiment analysis is used to extract features from the customer reviews which help enhance the performance of the selected predictive models. Feature importance analysis is also used to select the most representative features for predicting the price of the listings. Experimentation shows that SVR model can achieve an R-2 score of 69% and a MSE of 0.147 (defined on ln(price)) on the test set, outperforming the other models considered in the paper. [Link to the repository: github.com/PouyaREZ/AirBnbPricePrediction].