Research On Predicting Score Line Of College Entrance Examination For Sino-foreign Cooperative Education Based On Artificial Neural Network

Sino-foreign cooperation in running schools has developed rapidly and become an important form of modern education.From the perspective of "low frequency and high risk" of college entrance examination,how to predict the admission score line of Sino-foreign cooperatively-run college entrance examination effectively and accurately has become the most concerned topic for the candidates and parents who want to apply for such major.However,the uneven data of the fractional line and the numerous enrollment situations of various majors make the predicted results inaccurate.In addition,the examination institutions and prediction services mostly stay at the level of manual consultation,which leads to the problems of low utilization rate of test scores and inaccurate filling in.Therefore,based on the artificial neural network model and on the analysis of Sino-foreign cooperatively-run specialties,this paper introduces a comparative experimental group of similar specialties to explore the impact of joining similar specialties on the prediction of the score line of Sino-foreign cooperatively-run specialties.Firstly,this paper collects the professional fractional line data from 2015 to 2017,cleans and standardizes the data,codes the features,divides the data set according to the year,and takes the data of 2015 and 2016 as training set,and then extracts 20% of the samples as validation set in training set,and takes the data of 2017 as test set.Secondly,an artificial neural network model is constructed,including network node design,network parameter setting,training rate,iteration times,excitation function selection,etc.Then the artificial neural network model is optimized,including the selection of hidden layer nodes and the use of regularization method to prevent over-fitting of the network.Finally,the model is solved,and the prediction effect of joining the same profession is compared.The experimental results show that,with the same model,that is,the training topology,the number of iterations,the excitation function and other network topologies are unchanged;the predicted values of the artificial neural network after training are similar to the original values.The effect is more prominent,indicating that its accuracy is higher.By analyzing the error of the prediction model,it is found that the network prediction model after adding the same professional training is reduced by 9 points on average compared with the network prediction model without similar professional training.It can prove the network structure of this paper.And the selection of network parameters is more reasonable,and can play a certain reference role in the prediction of professional scores in most cases.According to the total number of samples distributed by each batch,the average error without the prediction of the same kind of professional is 2.81%,and the average error of adding similar professional predictions is 1.02%,which indicates that joining the similar profession can effectively improve the college entrance examination for Chinese-foreign cooperative education.The accuracy of the score prediction,which helps to improve the approximation ability of the function,and its accuracy,is improved by 1.79% compared with the traditional prediction.In this paper,some attempts have been made to apply artificial neural network to fractional line prediction.Preliminary experimental results show that this attempt is valuable,but there are also many shortcomings.Consider adding comparable specialty to other specialty prediction in the future,and consider using multi-parameter fusion and multi-activation to improve the neural network in order to improve the accuracy of prediction and hopefully discover new regularity.