Research On The Prediction Algorithm Of College Entrance Examination Score Line Based On Deep Learning

With the vigorous development of Chinese society,people pay more and more attention to the education of their children.The college entrance examination,as a national examination for the country to screen talents,has attracted more and more attention from the society.The scientific prediction of the college entrance examination is of great significance to the voluntary filling of the college entrance examination.This paper collects the scores of college entrance examination colleges and universities in Hubei Province from 2016 to 2020,and cleans the data,standardizes the numerical features,encodes the character features,and then selects 2016 to 2019 as the training set.The 2020 data is used as the test set,and then 30% of the training set is selected as the validation set.In this paper,the information fusion least squares method is first used for prediction.This method utilizes the principle that the score features and ranking features in the data set can be converted to each other.The two features are used for linear prediction,and then the results are compared and corrected to achieve the purpose of reducing errors.Compared with the direct regression prediction algorithm,the accuracy of the algorithm is improved by 3%.Secondly,this paper proposes a prediction model based on the orthogonal subspace direct sum decomposition algorithm.Ordinary Least Squares(OLS)approximates the vector using an orthogonal projection of the vector on the subspace,generates regression coefficients,and then computes the predicted vector while maintaining a linear relationship on the higher-dimensional subspace.In general,the predicted vector is no longer an orthographic projection of the approximated vector.Based on the direct sum decomposition principle of the orthogonal subspace,the orthogonal complementary subspace of the projected subspace is constructed,the error vector in the complementary subspace is calculated,the original vector is reconstructed,and the stability of the improved method is also analyzed.The experimental results show that the improved method has higher prediction accuracy than OLS,and the accuracy is improved by 5%.Finally,considering the nonlinear characteristics of the college entrance examination data,this paper adopts the LSTM neural network in the deep learning theory,and introduces the attention mechanism to optimize and improve it,and construct the LSTM-ATTE neural network model,which includes the design of network layers,network Layer parameter design,number of iterations,activation function,optimization function,etc.Finally,the model is verified by experiments.The experimental results show that the accuracy of the LSTM-ATTE neural network model can reach 93%.This paper attempts to apply deep learning theory to the prediction of college entrance examination scores.The experiments show that this attempt has research value.However,there are still many shortcomings.For example,the data set needs to be continuously improved and features are emphasized,so that the model operation efficiency can be improved.A variety of neural network model sets can be used to improve the accuracy of the model and have the opportunity to discover new rules..