Predictive Business Analytics using Different Predicting Model, Comparison of Decision Tree and Random Forest

Abstract

Predictive Analytics is a crucial branch for data-driving analysis in business, and could be used in cutomer behavior prediction and evaluation. The selection of an appropriate model involves a trade-off between prediction accuracy, interpretability and computational costs. This study aims to empirically compare the predictive performance of Decision Tree (DT) and Random Forest (RF) in identifying potential customer type and bahavior for a new travel package, addressing a challenge of model selection in realbusiness context. Using a customer dataset from a tourism company (“Visit with us”), it implemented and tuned both DT and RF models. Model performance was evaluated on key metrics including accuracy, precision, recall, and F1-score. The analysis found that tuned Random Forest Model is found to be the superior model with the highest test accuracy and F1 score, indicating a balance between precision and recall. While tuned Decision Tree Model ranked at the second. The study confirms the established superiority of ensemble methods like Random Forest for prediction tasks but provides a nauced business insight: the choice between DT and RF might depend on the strategic goal—maximizing customer reach versus optimizing marketing efficiency. The findings offer actionable strategies for targeted marketing and demonstrate the significant value of predictive analytics in formulating business strategy.