Estimating Individual Customer Lifetime Values with R: The CLVTools Package
This work proposes CLVTools, an open-source R package for customer lifetime value (CLV) modeling that addresses key challenges such as sparse transaction data and prediction horizons exceeding the observation window. Built upon probabilistic generative models—including Pareto/NBD and Gamma-Gamma—the toolkit integrates maximum likelihood estimation with Bayesian inference, and supports both time-invariant and time-varying covariates, parameter regularization, and equality constraints. Designed for robustness and computational efficiency, CLVTools delivers accurate individual-level CLV predictions even with limited data, while maintaining scalability to large datasets. By enhancing both predictive precision and data frugality, the package offers a practical and extensible solution for marketing decision-making.