This project explores Airbnb listings data from Athens for the years 2019 and 2023. The goal was to analyze trends in availability, pricing, reviews, and other key features over time, identify outliers, and extract meaningful insights about the short-term rental market.
The analysis combines data cleaning and transformation, statistical outlier detection (box plots, IQR filtering) and natural language processing (NLP) techniques (tokenization, stopword removal, bigram frequency analysis) to identify patterns in listing descriptions.
As part of the project, we also implemented a content-based recommendation system that suggests similar Airbnb listings based on text descriptions and key listing features, giving users a way to discover relevant alternatives.
This project builds on our previous analysis of Airbnb listings in Athens, focusing on guest review comments and their sentiment.
Our goal was to study how guest sentiment evolved between 2019 and 2023 and uncover patterns that could provide insights for hosts.
We applied several text preprocessing techniques and experimented with both TF–IDF and Word2Vec representations. Reviews were classified into positive and negative categories using multiple machine learning algorithms, including Support Vector Machines (SVM), Random Forests, and K-Nearest Neighbors (KNN), and we compared their performance to identify the most effective approach.