Figure 3
A decision tree diagram illustrating the predictive features of smoking behavior based on reverse marketing categories. The tree starts with the root node labeled Q4.4 Place, which splits into two branches: Accessible and Isolated. The Accessible branch leads to Q3.1 Price, which further splits into High price and Low price. High price leads to Q2.1 Product, which splits into Supply decreases and Supply stable, resulting in Non-smoker Likely to quit and Smoker Persists, respectively. Low price leads to Q3.4 Price, which splits into Fee increases and No fee change, resulting in Non-smoker Likely to quit and Smoker Persists, respectively. The Isolated branch leads to Q2.2 Product, which splits into Fewer cigarettes and No change in supply, leading to Q4.3 Place. Q4.3 Place splits into Ban increases and No ban, resulting in Non-smoker Reduced access and Smoker Persists, respectively.Simplified decision tree derived from random forest classification – top predictive features of smoking behavior by reverse marketing category (place, price and product)