Design Of College Admission Score Prediction Scheme Under The New 6 Pick 3 National College Entrance Examination Policy

The quality of Application Form of the College Entrance Examination depends on appropriate mark gradient between colleges and majors in the form.And this requires accurate prediction of College Admission Score.At present,mature prediction methods include Equivalence Score,Average Rank,Percentile Regression and so on.In September2014,the State Council issued the "Implementation Opinions on Deepening the Reform of the Examination Enrollment System",announcing the beginning of the New College Entrance Examination Reform.In 2020,“Arts or Science” College Entrance Examination is replaced by "6 choose 3" mode in Shandong.And the “S/R Mapping Table(Score Rank Mapping Table)” for Arts or Science mode was completely invalid under the “6 choose 3”mode,which led to the failure of the traditional College Entrance Examination Score Prediction Method.Although someone developed “Arts or Science Equivalence Score”method to predict college admission score,invalid prediction appears when Science proportion is too small in history data.In order to obtain more accurate prediction under first year “6 chooses 3” mode,a“Weighted Average Regression” method is developed.It depends on a virtual “6 chooses 3”dataset which is constructed by merging Arts and Science historical data,and then linear regression method can be used to predict the college admission scores.First,Percentile Normalization Method is used to normalize the “Arts or Science” mode historical college admission scores,and then admission plan is used as merging weight to obtain Colleges Virtual Historical Percentiles under “6 choose 3” mode.and after that,use linear regression model to predict the first year percentiles of college admissions,finally Spline Regression Method is introduced to map continuous percentile to score.Spline Regression Method can greatly reduce the amount of data in the “S/R Mapping Table”.By comparing prediction result of “Weighted Average Regression” and “Arts or Science Equivalence Score” method with Shandong actual college admission scores in 2020,“Weighted Average Regression” method has higher prediction Accuracy,and will not appear invalid prediction result even Science proportion is very small.A “P/S Mapping Table(percentile-score mapping table)” was constructed based on the “S/R Mapping Table”.More than 700 sets of “P/S Mapping” data were compressed to 20 sets of key-value pairs,and also achieves continuous P/S Mapping.By comparing the predicted scores of more than 1,000 colleges with the actual admission scores,“Weighted Average Regression” method average prediction error is 1.77%.Figure 15;Table 23;Reference 42...