Predicting Customer Booking Behaviour

Project Title: Predicting Customer Booking Behaviour

Tools and Libraries used: Python, jupyter notebook, pandas, ydata_profiling,matplotlib,seaborn,scikit-learn,imblearn,XGBoost

GitHub Repository: The full analysis code can be found here- Link

Approach:

• Loaded customer booking dataset using pandas.

• Generated data profiling report with ydata_profiling - Link.

• Selected relevant features and split data into train and test sets.

• Trained Logistic Regression model for feature importance.

• Evaluated XGBoost classifier model for accuracy.

• Visualized feature importances using bar plot.

• Identified top 5 important features.

Results:

• The XGBoost model attains an accuracy of 0.8512, which is 85%.

• Features positively correlated with booking completion - Extra baggage, preferred seat, in-flight meals and flight hour.

• Features negatively correlated with booking purchase - Purchase lead, length of stay and flight duration.

Visualization: Plot of Feature importance's

Conclusion:

• Passenger preferences for baggage, seating, in-flight meals, flight timing, and purchase lead time significantly impact booking decisions.