kk289 / Airbnb-Price-Prediction

A Survey on ML Techniques for Airbnb Price Prediction

STAT-493: Statistical Modeling

Senior Data Science Class, SPRING'20

Title: A Survey on Machine Learning Techniques for Airbnb Price Prediction

Author: Kevil Khadka

Abstract:

In this project, I develop a model to predict the Airbnb rental Price of Los Angeles, California (USA) based on their listing dataset, and also, I work on text analysis and sentiment analysis based on Airbnb's review dataset. This model is very helpful to find a price difference based on different characteristics like room type, minimum nights staying, neighborhood area, etc. Plus, it helps tourists to find out the best destination and right place to stay based on a review analysis of host listing. The dataset I used was provided by Airbnb's website called Inside Airbnb. The problem is modeled as a classification problem and found random forest method and naïve Bayes to be a very best model to predict the price based on the accuracy while multiple linear regression has the smallest Test MSE comparing to other models. The purpose of the project is also to find the best models based on accuracy and test mean square error (MSE). I use the following methods: Multiple Linear Regression, Decision Tree, Naive Bayes, Random Forest, Bagging, Linear Discriminant Analysis, Quadratic Discriminant Analysis, and k-Nearest Neighbor. For the text analysis and sentiment analysis, I use the review datasets for the following cities: Los Angeles, Chicago, New York, Boston, London, and Greater Manchester. I use the concept of Term frequency-inverse document frequency (tf - idf) to find out how important a word or comment is to a document or in a collection of documents, which focuses the difference between the reviews of the two cities.

Project Report

Datasets Links:

1. Los Angeles Listing Dataset: http://data.insideairbnb.com/united-states/ca/los-angeles/2018-12-06/visualisations/listings.csv

2. Los Angeles Review Dataset: http://data.insideairbnb.com/united-states/ca/los-angeles/2018-12-06/visualisations/reviews.csv

3. Chicago Review Dataset: http://data.insideairbnb.com/united-states/il/chicago/2018-12-13/visualisations/reviews.csv

4. London Review Dataset: http://data.insideairbnb.com/united-kingdom/england/london/2018-12-07/visualisations/reviews.csv

5. Greater Manchester Review Dataset: http://data.insideairbnb.com/united-kingdom/england/greater-manchester/2018-12-10/visualisations/reviews.csv

6. New York Review Dataset: http://data.insideairbnb.com/united-states/ny/new-york-city/2018-12-06/visualisations/reviews.csv

7. Boston Review Dataset: http://data.insideairbnb.com/united-states/ma/boston/2018-12-13/visualisations/reviews.csv