Improved LSTM with Attention Mechanism for Large-Scale Educational Data Analysis and Teaching Strategy Optimization

In the research on innovative teaching strategies and practices of vocational education, the traditional LSTM model has high computational complexity and low efficiency when analyzing large-scale educational data, and the lack of continuous feedback and optimization mechanism leads to lagging teaching strategies. This paper constructs an improved model of LSTM (Long Short-Term Memory), performs preprocessing operations such as cleaning, denoising, and normalization on the collected large-scale education data, and extracts key features to provide high-quality input for subsequent model training; the improved LSTM model introduces an attention mechanism optimization strategy to reduce computational complexity and improve the model’s prediction ability for time series data. Based on the prediction results of the improved LSTM model, it designs a real-time feedback system for monitoring students’ learning status and adjusting their teaching. A teaching strategy optimization and iteration mechanism is established to regularly review and adjust teaching strategies based on the model’s teaching effects, forming a closed-loop optimization process to continuously improve teaching quality. The experimental results show that the RMSE (Root-Mean-Square Error), MAE (Mean Absolute Error), MAPE (Mean Absolute Percentage Error) and R 2 of the improved LSTM model in predicting students’ learning status are 35%, 30%, 5% and 85% respectively. When processing 5000 data, the computing time and memory consumption were 6.8 seconds and 3.0 GB respectively, which showed better model performance. In the dynamic adjustment of teaching strategy innovation, the average response time was 3.15 seconds, and the students’ learning status was significantly improved after the adjustment. The experimental results proved the effectiveness of this study.