Mathematical Achievement Prediction Based On Machine Learning

Mathematics education is an important part of the basic education,and improving students' mathematics level is an important driving force for the sustainable development of mathematics education.The mathematics achievement reflects the mathematics level of the students to a certain extent,and its rationality can not be ignored because of the limitation of evaluating the students' mathematics level.With the strong support of the state for the development of education in the western regions,the basic education has made great progress,but it is still far from the Eastern Region in general.From the college entrance examination,there is a certain gap between the average scores of Tibetan students and those of the eastern region as a whole.In this paper,based on the machine learning prediction theory and technology of Educational data mining,we use the survey data of four high schools in Lhasa as the starting point,this paper studies the important factors which affect the mathematics achievement of Lhasa Students,and gives the explanation of the related research of educational psychology,and puts forward some suggestions to improve the mathematics achievement of Lhasa students.This research is mainly divided into three parts: Theoretical study,concrete realization,and conclusion and suggestion:(1)Research the learning Educational data mining,describe the processes and metrics associated with ML prediction systems,and analyze the typical classification algorithms involved in ML prediction systems.(2)In order to provide reference for the analysis and prediction of Lhasa data,the CEPS data are analyzed and predicted first.First of all,CEPS data pre-processing,including data acquisition,format conversion,data cleaning,data segmentation.Data cleaning includes data missing processing,redundancy deletion,outlier processing and feature transformation.Secondly,the data are divided into training set,test set,and modeling in K nearest neighbor,logistic regression,linear Support vector machine,Support vector machine,decision tree,XBG classifier,Ada classifier,GradientBoosting classifier,random forest,Gaussian Bayesian,Bernoulli Bayesian,polynomial Bayesian machine learning classification algorithms,to evaluate the prediction effect of each model.Finally,the parameters of the better prediction model are adjusted to determine the best model,and the important characteristics that affect the mathematical results are output.Based on the data from the China Education Panel Survey and the related research and design of the questionnaire,this paper investigates the current situation of mathematics learning of senior one students in four senior high schools in Lhasa,and collects the data.Firstly,the data were preprocessed and the data distribution was analyzed by using the feature visualization technology,and the correlation analysis was made on the grades of the subjects in grade one of the high school in Lhasa,as a supplement to the prediction results.Then,based on the distribution characteristics of the survey data,the stepwise regression analysis is used to classify and predict the students' mathematics achievement,and the output affects the important characteristics of the mathematics achievement.Finally,12 kinds of classification prediction models are established,and the model parameters are adjusted to determine the best classification model of mathematical achievement prediction.(3)Two conclusions can be drawn from the results of the analysis of the prediction model.First,students of different levels have significantly different sense of self-cognition of learning mathematics,which indicates that students' sense of learning is most related to their mathematics achievement.This is consistent with CEPS data analysis.Second,the native language has an important influence on the mathematical hierarchy of students.This is consistent with the results of 32 Variables Correlation Analysis,Feature Visualization Histogram,correlation analysis of students' scores and stepwise regression principal component analysis.In order to improve academic performance,teachers should reward students in time,increase students' sense of achievement,enhance students' self-confidence in learning and enhance students' sense of self-efficacy.At the same time,teachers should pay attention to the cultural background of students' native language while teaching,and educational design should consider the influence of language differences on students' cognition.