In the context of fluctuating international oil prices, an in-depth exploration was conducted to investigate the impact of media indices, macroeconomic variables, and selected technical indicators on the Chinese stock market. By constructing two prediction models, UMAP-LSTM and ISOMAP-GBR, the daily stock prices of 10 listed companies from different industries in China were successfully and accurately predicted. The research results indicate that both models demonstrate high predictive accuracy, and the incorporation of media indices into the models resulted in statistically superior prediction performance compared to models that did not consider media factors. This finding emphasizes the central role of media-related information in explaining stock market fluctuations. Further analysis reveals that global and Chinese media hype indices have a significant impact on stock price changes, with increased media hype potentially leading to market declines, reflecting the important guiding role of media sentiment on investor behavior and market trends. Additionally, the interaction between media indices also has a significant effect on stock prices, further confirming the crucial role of media sentiment in driving stock market volatility. The research conclusions not only provide investors with more accurate stock market prediction tools but also reveal the important role of media sentiment in financial markets, offering valuable references and decision-making bases for policymakers, market regulators, and investors, and contributing to a deeper understanding and response to financial market fluctuations.