LA Airbnb Analysis

An analysis and Tableau dashboard for LA Airbnb host performance, pricing, listings, and review sentiment.
Author

Khang Thai

Published

May 29, 2024

R Random Forest Tableau Tidyverse Caret

Introduction

The LA Airbnb Analysis project analyzes Airbnb listings and reviews in Los Angeles. The objective is to provide insights and strategies that help hosts improve performance and profitability. The project uses listings and review datasets, then focuses on factors that contribute to superhost status, pricing, sentiment, and listing visibility.

Process

  1. Host Analysis:
    • Analyzed key predictors such as response rates, acceptance rates, ratings, profile pictures, and duration of hosting.
    • Used logistic regression and random forest models to determine factors contributing to superhost status.
  2. Listing Analysis with NLP:
    • Applied natural language processing to conduct sentiment analysis of reviews.
    • Used Latent Dirichlet Allocation for topic modeling to identify themes in guest reviews.
  3. Dashboard Development:
    • Created an interactive dashboard displaying price, room type, beds, ratings, and location.
    • Included filters for guests to find listings based on their preferences.

Interactive Dashboard

If the embedded dashboard does not load, open it directly on Tableau Public.

Open Interactive Dashboard

Outcome

  • Key predictors for achieving superhost status included high ratings, acceptance rates, and response rates.
  • Test accuracies for logistic regression and random forest models were 74% and 76%, respectively.
  • Suggested strategies included maintaining high ratings, addressing negative reviews, and improving acceptance rates.
  • Sentiment analysis helped identify common issues with low-rated listings.