A Machine and Deep Learning Framework to Retain Customers Based on Their Lifetime Value

Customer Lifetime Value (CLV) measures the average revenue generated by a customer over the course of their association with the firm. The Recency Frequency Monetary (RFM) Model is used to calculate the CLV. Recency is the latest item purchased. The number of times an item is purchased is the Frequency. Monetary is the price spent on the product by customers. CLV is measured using previous customer transactions of RFM factors. This research proposes a Deep Learning Customer Retention Framework to predict the Customer Lifetime Value in order to retain customers through an effective Customer Relationship Management strategy. The proposed framework combines clustering and regression models to analyze the significant variables for predicting the lifetime value of customers. Customers are categorized into levels such as high medium and low profitable customers based on their lifetime value. This research compares Deep Neural Network models, Machine Learning models and Probabilistic models. The Deep Neural Network is ANN. The machine learning models are Linear Regression, Random Forest, Gradient Boosting. The probabilistic models are Gamma-Gamma and Betageometric/negative binomial. The models are compared in order to predict the level of profitable customers. Results demonstrate that Deep Neural Network (DNN) model outperforms the other models with 71% accuracy. Improved prediction model for CLV and segmentation assists the firms to plan and decide relevant CRM strategies such as customer profitability analysis, cross-selling and one to one marketing for the future.