New Teaching Model of Professional Big Data Courses in Universities Based on an Outcome-Oriented Educational Concept

Abstract

Most traditional Big Data courses focus on the cultivation of students’ professional theoretical knowledge but neglect the training of students’ programming skills. To test the effectiveness of students’ programming learning, the study constructed a classification algorithm based on the TCNN model from the perspective of program identification for identifying programs written by students on their own. To improve the convergence speed and to retain more data features of the programs, the study used the Softplus-Relu combined activation function. To avoid overfitting the model, a dropout strategy was also introduced to optimize the existing TCNN model. The experimental results show that the TCNN model with the Softplus-Relu activation function converges faster, and the classification accuracy obtained is higher than 95%. The loss values and classification accuracies obtained with the TCNN-Dropout model are better than those of the GIST-KNN algorithm. The former has a loss value close to 0 and an accuracy rate of 98.9%. Thus, this indicates that the improved TCNN model proposed in the study has advantages in the identification procedure and can be used as a teaching aid for the training of big data professionals.