Logistic Regression and Decision Tree
Let’s face it, the travelling public can be at best a fickle bunch and at worst, well…
Our model predicts customer cancellations by applying Logistic Regression and Decision Tree models to customer history and demographic data including lead time, booking type, day of the week, number of children and room price. The hotel can gain an edge by identifying customer characteristics, seasonal variabilities, room type and service level to adjust their cancellation policy and target rebooking.
EDA
Treatment for Multicollinearity and Outliers
Built Logit model to a p-value of 0.05 with a regression equation to interpret coefficients as odds
ROC-AUC to optimize model performance
Decision Tree on original data with Pre and Post Pruning.
