Predicting Consumer Behavior and Preferences in E-Commerce Using Convolutional RNN Techniques
Abstract
Worldwide trade has been utterly transformed by the meteoric rise of the internet and the digital economy in the last 20 years. The effectiveness of advertising products and services online has been greatly improved by the rise of digital platforms, high-speed internet connections, and web technologies. More and more people are choosing to shop online instead than at physical stores. More and more people are buying online because they have high expectations for the convenience and security offered by online stores. Data preparation, feature selection, and training of the ConvRNN Model are presented in this study. Following the integration and purification of data during preprocessing, PCA is used to choose characteristics according to customer purchasing behaviour and preferences. The ConvRNN technique uses CNNs to track regional buying habits, and LSTMs to improve forecast accuracy by capturing long-term temporal correlations. This method was able to identify high-margin e-commerce opportunities by identifying crucial periods just before large stock performance variations. The 96.28% accuracy rate of the ConvRNN model is higher than that of conventional approaches. In order to improve inventory strategies, hone customer engagement, and enable data-informed market performance decisions, these studies show that AI-driven models can predict e-commerce consumer buying behaviour and preferences.