Exploring Consumer Purchase Psychology in E-Commerce Using BiLSTM and Dense Neural Network Models

Modern e-commerce platforms provide a large quantity of products, but the sheer volume of information can be daunting to users. The ability for product suggestion algorithms to cater to specific users' tastes greatly improves their overall experience. Since consumer spending habits reflect fundamental psychological processes, tailored ideas boost involvement and contentment. Using a proprietary recommendation system, this study examines consumer behaviour on local e-commerce sites that facilitate group purchase. Applying NLP to analyse data provided by users enhances automation and information retrieval. In order to uncover important details, natural language processing looks at social structures. We reduce dimensionality using a principal component analysis method that employs PEVD. To predict how much money consumers will spend, we use Dense-BiLSTM. When it came to forecasting customers' buying habits, the Dense-BiLSTM model hit a validation accuracy of 95.58%. Natural language processing and dimensionality reduction improve the accuracy and efficiency of recommendations. The results show that customised recommendations are much improved with the use of strong machine learning algorithms and NLP analysis. By reducing data overload, the suggested method improves the efficiency and customisation of online shopping. Future studies might refine the model to account for customers' changing behaviour in real time.