Data Analysis of Product Information and Reviews on E-commerce Platforms Based on Python

In the digital environment, e-commerce platforms have accumulated a large amount of both structured and unstructured data. As a core carrier of user perception, review content has a significant impact on consumer behavior and platform operations. To systematically mine such information, this study constructs a multi-level analytical framework. First, based on Python, multi-source review data is subjected to sentiment annotation and visual analysis to reveal basic distribution patterns. Furthermore, for the small-sample scenario, a hybrid 1-AGO-Prophet model is employed to explore the feasibility of trend forecasting for review volume; simultaneously, the XGBoost model is applied for fine-grained sentiment classification, with a focus on examining the impact of class imbalance issues. Experimental results show that the XGBoost model achieves high accuracy in recognizing positive sentiment but exhibits limitations in identifying sparse negative reviews. In contrast, the forecasts of the Prophet model exhibit considerable uncertainty due to the limited sample size. This study not only verifies the applicability of the relevant methods in e-commerce review analysis but also delves deeper into the challenges posed by data scarcity and class imbalance, providing a reference for platforms to intelligently manage reviews that balance methodological innovation with an awareness of limitations.