Evaluating Deep Learning Models for Customer Lifetime Value Forecasting Based on Hybrid DT and Naive Bayes Model

Customer lifetime value, a core concept in direct response marketing for quite some time, is now receiving increasing attention from management at Boston University's School of Management. However, the majority of published works on the topic have mainly concentrated on lauding its usefulness as a decision-making metric, offering only one numerical example of how it is calculated or determined, incorporating it into broader discussions of profitability, and examining its function in decisions regarding customer acquisition and the tensions between acquiring and retaining customers. There are three steps to the suggested method: preprocessing, feature extraction, and model training. After archiving duplicate and missing data, data preparation is employed to eradicate missing values. The feature selection process incorporates K-means, principal component analysis (PCA), and customer lifetime value measurements. An increase in the model training procedure's accuracy was achieved by utilizing the Hybrid DT and naive bayes. The recommended approach achieves a higher accuracy rate of 96.03 % compared to its rivals, which includes naive bayes and Hybrid DT.