Airbnb Listing Price Prediction in London

This repository contains the code for my project, where I predict the listing prices of Airbnb properties in London by leveraging both textual and tabular data. The goal of this project was to develop a robust machine learning pipeline capable of accurately estimating prices using advanced models and a stacked ensemble approach. The data was sourced from a kaggle competition.

Project Overview

The project follows a structured approach:

1. Exploratory Data Analysis (EDA):
   A thorough examination of the dataset to identify features with strong predictive power and understand patterns in the data.

2. Data Cleaning and Preprocessing:
   Includes handling missing values, encoding categorical variables, and preparing text and tabular features for modeling.

3. Model Building and Evaluation:
   Various models were tested and evaluated to identify the best-performing ones:

   For Tabular Data:

   • Linear Regression
   • Random Forest
   • XGBoost
   • Deep Neural Network

   For Text Data:

   • LSTM
   • GRU
   • Bidirectional GRU
   • BERT

4. Stacked Ensemble Model:
   The final solution employs a stacked ensemble consisting of:

   • XGBoost for tabular data.
   • BERT for textual data.
   • A DNN to merge the predictions from XGBoost and BERT into a final output.