The prediction of stock prices has garnered increasing attention from the data mining and machine learning communities. Accurate forecasting results can assist investors in mitigating investment risks. Due to the volatility of the stock market and policy influences, stock data also exhibits high levels of volatility and randomness, thereby often aligning with market sentiment. Moreover, various original stock-related datasets encompass ample historical information that can be utilized to analyze future trends; however, traditional forecasting models struggle to effectively leverage this information, limiting their learning capabilities and reducing prediction accuracy. This paper presents a novel stock prediction model that integrates multi-perspective features derived from stock data with ensemble learning techniques. By combining both stock index data and sentiment analysis data within an integrated framework, our model effectively captures the correlation between the target stock index and market sentiment orientation for accurate predictions. Experimental evaluations conducted on three prominent indices (CSI 300 Index, SSE 50 Index, CSI 500 Index) demonstrate outstanding accuracy and returns achieved by our proposed model.