Deep learning-driven customer segmentation model for online retail industry based on extended RFM model

In recent years, customer segmentation has become a crucial tool in the retail industry. Customer groups are segmented based on traits to create customized marketing campaigns. Customer segmentation approaches overlook complex hidden patterns in consumer datasets. Traditional RFM (Recency, Frequency, and Monetary) analyses segment customers but lack depth in defining customer dynamics. To resolve these issues, research in deep learning (DL) and machine learning (ML) is used to develop a customer segmentation and prediction process for the retail industry. This study develops an endwise predictive model for customer segmentation with purchase behavior Analysis using an Extended RFM with Fine-Tuned Convolutional DL (EXRFM-FTCDL) model. Segmentation and prediction are two stages of the EXRFM-FTCDL model. EXRFM features with behavioral measures (recency, frequency, money, repurchase time, average order value (AOV), and interpurchase time) are derived during segmentation, and a Gaussian Mixture Model (GMM) is applied to cluster. EXRFM-FTCDL model chooses the optimum number of clusters exploiting the Bayesian Information Criterion (BIC) and the Silhouette score. An improved artificial rabbit optimization (IARO) method using a hybrid 1-dimensional convolutional long short-term memory approach (Conv1D-LSTM) network has been used in the prediction stage of the EXRFM-FTCDL method. A retail benchmark dataset from a benchmark repository is employed for performance evaluation of the EXRFM-FTCDL technique. According to the results of the experiments on the public Online Retail dataset, the EXRFM-FTCDL methodology is effective at segmenting and classifying regular, occasional, and high-frequency customers. According to a comprehensive study comparing EXRFM-FTCDL methodology to other state-of-the-art models, it had better evaluation metrics.