🏠 Airbnb Data Science: Sentiment Analysis & Price Prediction

This project addresses two main challenges using Airbnb data:

1. Sentiment Analysis: Classifying Airbnb reviews as positive or negative to help hosts, guests, and platforms gain insights into user experiences.
2. Price Prediction: Predicting the nightly price of Airbnb listings based on features such as location, amenities, and review sentiment.

---

📊 Project Overview

The project is divided into two main parts:

1. Sentiment Analysis

We leverage transfer learning by pretraining models on publicly available datasets:

• IMDB Movie Reviews – clear polarity-labeled text data

This approach enables better generalization and more effective sentiment classification on Airbnb reviews.

2. Price Prediction

We use machine learning regression models to predict Airbnb listing prices. Features include:

• Listing characteristics (location, number of rooms, amenities, etc.)
• Aggregated sentiment scores from guest reviews

This enables more accurate and data-driven pricing strategies for hosts and platforms.

---

⚙️ Methodology

• Data Preprocessing: Cleaning and transforming text and tabular data.
• Model Training:
  • Sentiment Analysis: Models like SGDClassifier, Logistic Regression, etc., trained on IMDB and fine-tuned on Airbnb.
  • Price Prediction: Regression models (e.g., Linear Regression, Random Forest) trained on Airbnb listing data.
• Evaluation: Using metrics such as accuracy, F1-score (for sentiment) and RMSE, MAE (for price prediction).

---

🛠️ Technologies Used

---

📁 Project Structure

• Price Prediction/ – Contains the files and notebook to predict nightly prices based on features
• Sentiment Analysis/ – Contains the files and notebook to classify airbnb reviews between positive and negative sentiments

---

📽️ Demo

---

👥 Authors & Collaborators

This project is being developed by:

• Unai Lalana
• Filippo Muscherà
• Eneko Isturitz
• Paul Gasnier