Abstract:

Intelligent technology development is gaining traction in the sphere of education. The increasing rise of educational data suggests that standard processing techniques may be limited and distorted. As a result, recreating data mining research technology in the education area has become more important. To prevent erroneous assessment findings and to anticipate students' future performance, this research analyses and predicts students' academic achievement using applicable clustering, discriminating, and convolution neural network theories. To begin, this work suggests that the clustering-number determination be optimised by using a statistic that has never been employed in the K-means approach. The clustering impact of the K-means method is next assessed using discriminant analysis. The convolutional neural network is presented for training and testing with labelled data. The produced model may be used to forecast future performance. Finally, the efficacy of the constructed model is tested using two metrics in two crossvalidation procedures in order to verify the prediction findings. The experimental findings show that the statistic not only addresses the objective and quantitative problem of determining the clustering number in the K-means method, but also enhances the predictability of the outcomes.

Description:

Academic performance, clustering analysis, convolutional neural networks, discriminant analysis, educational data mining