British Airways — Customer Booking Analysis

Airlines collect a large volume of booking data — route, timing, passenger details, extras selected — but converting that into a usable signal for which bookings are likely to complete (versus abandoned) requires cleaning, exploration, and a model that can rank feature importance in a way a business team can actually act on.

The project follows a standard applied data science pipeline: clean and preprocess the raw booking dataset, run exploratory data analysis to understand distributions and correlations, engineer features from the raw booking attributes, and train a Random Forest classifier to predict booking completion.

Pandas for cleaning and preprocessing, exploratory analysis and visualization with Matplotlib, feature engineering on categorical and numerical booking attributes, and a Scikit-learn Random Forest classifier evaluated with standard classification metrics and feature importance ranking.

TODO: add the actual accuracy/precision/recall numbers, the top features by importance, and the final chart(s) once I pull them back out of the original notebook.

The cleaning and feature engineering stage took longer than the modelling stage — which is fairly typical, but easy to underestimate going in.