Customer LTV Prediction with RFM Model

★ 7.2 · general

ltv-predictor is a Claude Code skill that predicts customer lifetime value (LTV) using RFM feature engineering and regression models. It automatically computes Recency, Frequency, and Monetary metrics from order history, trains both linear regression and random forest models, compares them by R², and returns predictions with confidence intervals. On a benchmark dataset of 37,060 orders the random forest reached R² = 0.59 versus 0.48 for linear regression, with the Monetary feature accounting for 78% of importance. Reports are exported as HTML, Markdown, or Excel files; the skill ships with five core modules — data_processor.py, ltv_predictor.py, regression_models.py, visualizer.py, and report_generator.py — plus example scripts and sample data. It is aimed at e-commerce and retail teams that need to segment customers by predicted value and allocate acquisition or retention budgets more efficiently.