A Competitive Product Identification Model for FMCG Based on Siamese Neural Network

In the highly competitive fast-moving consumer goods (FMCG) market, accurate identification of competitive products is essential for enterprises to optimize product positioning, market segmentation, and brand strategy. Traditional competitor identification methods struggle to handle complex, high-dimensional, and heterogeneous data environments. To address these limitations, this study proposes a novel competitor identification model based on a Siamese Neural Network (SNN), which integrates both intrinsic product attributes and external consumer environment data. The model constructs pairwise comparisons of products and learns a distance-based similarity function through shared-weight neural architectures. Experimental evaluations on real-world FMCG datasets demonstrate that the proposed model significantly outperforms traditional machine learning baselines such as XGBoost, LightGBM, and Random Forests, achieving an accuracy of 86.2% and an F1-score of 84.8%.