Deep Learning for Enhanced Consumer Behavior Analysis and Predictive Accuracy
Jigyasa Bharti, Snehlata Dongre · 2024
The accelerating digital transformation in retail necessitates advanced analytical approaches to effectively comprehend and predict consumer behavior. Existing methodologies in retail data mining often fall short in handling the complexity and heterogeneity of data, resulting in suboptimal predictive accuracy and operational efficiency. This work addresses these limitations by introducing a novel deep learning framework that synergizes the capabilities of Graph Neural Networks (GNNs) with Q Learning and VARMAx models. The proposed model adeptly fuses data from diverse sources such as Twitter, Amazon, Facebook, and Flipkart, creating a comprehensive analysis platform. By leveraging the relational data structure processing power of GNNs and the dynamic decision-making proficiency of Q Learning, along with the time series predictive strength of VARMAx models, this approach offers a more holistic understanding of consumer trends. The efficacy of this model is demonstrated through extensive testing on varied data samples encompassing social media, e-commerce platforms, and tweets. The results are compelling: an 8.3% increase in accuracy, 8.5% enhancement in precision, 6.5% improvement in recall, and a significant 10.4% reduction in delay compared to existing classification methods. The impact of this work is far-reaching, offering retailers a robust tool for strategic decision-making, enhanced customer insight, and a competitive edge in the rapidly evolving digital marketplace. This approach not only propels the retail industry forward but also sets a new benchmark in the application of deep learning for consumer behavior analysis.