Sentiment Analysis and Predictive Modelling of Airline Customer Reviews

This project involves scraping customer reviews for British Airways from AirlineQuality.com, performing data cleaning and preprocessing, and conducting sentiment analysis using Natural Language Processing (NLP) techniques. The reviews are analyzed to classify sentiments as Positive, Negative, or Neutral using the VADER sentiment analysis tools. Additionally, the project includes exploratory data analysis (EDA) on customer booking behavior, where various features are analyzed to predict whether a booking is completed or not. The predictive model is built using a Random Forest Classifier, and the feature importance is evaluated using Mutual Information Scores. The project also includes data visualizations techniques such as pie charts and word clouds to present the findings effectively.

Model Performance: The Random Forest Classifier achieved a high training accuracy score 99.98% and a validation accuracy score 85.39% when using all features. This indicates that the model is effective in predicting booking completion, although the training accuracy suggests potential overfitting. The model's performance on the validation set demonstrates its ability to generalize well to unseen data.

Overall, the project provided a comprehensive understanding of customer sentiments and booking behaviors, offering actionable insights for improving services and enhancing customer experiences at British Airways.