Prediction Model Based On Association Rules And RBF Neural Network

With the rapid development of Internet and computers and the improvement of the awareness of information promoting teaching,in the field of education,more researchers are aware that the application of data analysis technology in improving the potential of teaching and learning,the progress of learning analysis technology is developing at a new speed in the day.In recent years,student prediction has gradually become an important part of learning analysis.Many researchers have constructed different learning prediction models from different perspectives based on different educational environments.All kinds of prediction models have their own characteristics,but they also have limitations in adaptability and generalization.The prediction model limits the specific input and parameter setting to prevent the promotion of the model.Therefore,a model with wide adaptability can make it an urgent need for educational researchers to predict from a variety of angles according to the characteristics of their own educational environment.Based on the analysis of the existing learning prediction model,based on the basic framework of the prediction model,the design pattern of the optimized RBF network prediction model based on the association rules mining and the related data preprocessing method are proposed by using the neural network technology and the association rule mining technology.First,we explain the method of converting numerical data into transactional data set data which can be analyzed by association rules.Then,the design model of the model is explained,and the RBF neuron data center selection optimization algorithm based on the support degree and confidence of association rules is proposed.The algorithm is different from the traditional method of random selection of neural data centers,and the prediction data and results have been known.On the basis of the correlation,the selection method of the data center is changed to the probability of random selection.This algorithm can speed up the learning speed of the network and improve the prediction performance of the network to a certain extent.Aiming at the sparse characteristics of offline educational data,based on previous studies,this paper proposes interpolation method for abundant sparse data.In order to increase training data,the purpose of improving network training performance is achieved.Finally,this model is applied to two practical education forecasts,the results of a secondary school student and the prediction of the employment of a certain undergraduate,and the comparison of the difference in the rate and accuracy of the learning rate and the accuracy of the optimized and not optimized neural network algorithm by the method of the control group experiment.The prediction model in this paper is compared with other mainstream prediction models to further evaluate this model.The results show that the prediction model proposed in this paper can improve the application range of the model,and improve the prediction efficiency and accuracy of the model,which has good performance and strong practical significance.It also provides a new way of forecasting models for educators in the field of education.