After the COVID-19 ended, the global economy gradually recovered. Due to the nonlinearity, complexity, and high noise of financial time series, stock price prediction has become one of the most challenging tasks in the stock market. To tackle this challenge and enhance the prediction performance in the complicated stock markets, we propose a novel integrated approach based on Complete Ensemble Empirical Mode Decomposition with Adaptive Noise (CEEMDAN), Long Short-Term Memory (LSTM), and ensemble learning algorithm LightGBM to simultaneously improve the fitting and accuracy of stock price prediction. In addition, to prevent overfitting and improve predictive performance, this study adopted the Simulated Annealing (SA) algorithm for optimization. The predictive performance of the proposed hybrid model is comprehensively evaluated by comparing it with single LSTM, RNN, and other popular hybrid models. Three evaluation metrics, namely Root Mean Square Error (RMSE), Mean Absolute Error (MAE), and accuracy, are used to compare the aforementioned models. The experimental results indicate that the proposed hybrid CEEMDAN-LSTM-SA-LightGBM model outperforms all other comparative models in this study with better fitting and accuracy.
As a key engineering discipline, the electronic information major faces an increasingly urgent demand for cultivating technically skilled talents. This paper addresses the current issues in electronic information major teaching by proposing a JiTT+CDIO teaching model, which combines Just-in-Time Teaching (JiTT) with the CDIO engineering education model. We design a teaching reform plan based on this model and discuss its specific implementation strategies in course teaching, practical components, and assessment through case analysis. The study shows that the JiTT+CDIO model can effectively improve students' learning enthusiasm, cultivate their engineering practical abilities and innovative capabilities, thus providing a new approach for the teaching reform of electronic information majors.
Stock market forecasting has always been one of the most difficult challenges faced by the stock market. In the issue of stock price prediction, its noisy nature, complexity, as well as the country's governance system, economic situation, epidemic will be mentioned. In recent years, scholars have conducted many stock market prediction studies using traditional and artificial intelligence technologies since the lower the error in stock prediction, the more investment risk can be reduced. In this research, a novel prediction method based on mode decomposition algorithm and optimized Support Vector Regression (SVR) is proposed. Mode decomposition algorithm is used to smooth the original signal and reduce the impact of noise. SVR extracts knowledge from data and ultimately utilizes the knowledge to predict new stock data. Bayesian optimization (BO) algorithm is adopted to optimize the core parameters of SVR and prevent overfitting so as to improve predictive performance. Finally, The CEEMDAN-BO-SVR model proposed in this study is compared with several other methods, including CEEMDAN-SVR, BO-SVR, SVR, RNN, LSTM, and ensemble learning algorithms, then analyzed through evaluation indicators. The experimental results comparing different models indicate that the CEEMDAN-BO-SVR model has the best fitting ability.