College file-casting line prediction method, system and device, and medium
By combining ARIMA and LSTM models, using the ARIMA model to extract the linear relationship of the college entrance examination filing line and using the LSTM model to process the error, the problem of insufficient prediction accuracy and stability in the existing technology is solved, and a higher prediction accuracy of the filing line is achieved.
Patent Information
- Application Number
- CN202410922962.6
- Authority / Receiving Office
- CN · China
- Patent Type
- Applications(China)
- Current Assignee / Owner
- Filing Date
- 2024-07-10
- Publication Date
- 2024-12-06
- Estimated Expiration
- Not applicable · inactive patent
AI Technical Summary
In the prior art, when the ARIMA model or LSTM model is used alone to predict the college entrance examination filing line, the accuracy and stability of the prediction results are poor, and it is impossible to effectively capture the linear and nonlinear relationships in the time series.
Combining the ARIMA model and the LSTM model, by standardizing the variables that affect the filing line in previous years, using the ARIMA model to extract linear relationships and make predictions, and using the LSTM model to process prediction errors, thereby improving prediction accuracy and stability.
Through the complete extraction of time series information by combined models, the prediction accuracy of the college entrance examination filing line is significantly improved, and it has higher prediction accuracy than using ARIMA or LSTM models alone.
Smart Images
Figure CN119090039A_ABST
Abstract
Description
Technical Field
[0001] The present invention relates to the field of data prediction technology, and in particular to a method, system, device and medium for predicting the admission score of a college. Background Art
[0002] As the most fair talent selection mechanism, the college entrance examination is not only an important means of evaluating students' academic level, but also a key indicator for various higher education institutions to select outstanding talents. Under this mechanism, the admission score becomes a key criterion for admissions to major universities, and has received widespread attention since the announcement of the college entrance examination results. Whether the admission score is overestimated or underestimated, it may cause candidates to lose the opportunity to enter their favorite universities, thus having a significant impact on their future education and career development. Therefore, how to accurately predict the admission score has become a major challenge that candidates must face in the process of filling out their applications.
[0003] At present, most of the predictions for the college entrance examination admission score are made using the ARIMA model or the LSTM model. The ARIMA model analyzes the linear relationship in the time series through stationarity test analysis and obtains the seasonal relationship in the series; the LSTM model captures the long-term dependency in the series by introducing a gating mechanism to complete the prediction of the time series. However, time series data contains both linear and nonlinear relationships. The ARIMA model cannot handle nonlinear relationships, and the LSTM model is not good at capturing linear trends and seasonal components in the model.
[0004] However, due to the complexity and diversity of college entrance examination data, the accuracy and stability of the prediction results of the ARIMA model or LSTM model alone for the admission score prediction are poor. Summary of the invention
[0005] In view of the shortcomings of the prior art that the accuracy and stability of the prediction results of the admission score are poor when the ARIMA model or the LSTM model is used alone to predict the admission score, the present invention proposes a method, system, device and medium for predicting the admission score of a college, which utilizes the linear prediction advantage of the ARIMA model to optimize the ability of the LSTM model to process time series data, thereby improving the accuracy and stability of the overall prediction model, thereby solving the problems existing in the prior art.
[0006] A method for predicting the admission score of a college or university comprises the following steps:
[0007] Select the variables that have affected the college admission score in previous years; the variables include the provincial control line and the total number of college entrance examination candidates as independent variables and the admission score position value as the dependent variable;
[0008] The independent variables and dependent variables are standardized to obtain an independent variable matrix M and a dependent variable column vector Y with time as rows and characteristic variables as columns, respectively;
[0009] Input the independent variable matrix M and the dependent variable column vector Y into the ARIMA model to obtain the predicted value M' of M and the predicted value Y' of Y, as well as the predicted value y' of the position value of the admission line in year T+1 T+1 ;
[0010] According to M and its predicted value M', Y and its predicted value Y', we get the error matrix ΔM of M and the error vector ΔY of Y, and at the same time, we get the autoregressive coefficient p of the ARIMA model of the dependent variable column vector Y; we use the matrix ΔX composed of the first T rows of data of ΔM as the independent variable and ΔY as the dependent variable to build the LSTM residual model;
[0011] The autoregressive coefficient p is used as the time step of the LSTM model, and the matrix composed of the last p rows of ΔM is input into the LSTM residual model to obtain the error value Δy of the position value of the admission line in year T+1. T+1 ;
[0012] Combined with the predicted value y' of the T+1 year admission line position value T+1 and error value Δy T+1 , obtain the predicted value of the college admission line in T+1 year.
[0013] Furthermore, the independent variables and dependent variables are standardized using the Min-Max standardization method to obtain the independent variable matrix M and the dependent variable column vector Y with behavior time and column as feature variables, which are respectively expressed as:
[0014]
[0015] Y=(y1,y2,…,y T ) T
[0016] Among them, X t =(x t1 ,x t2 ,…,x tN ) represents the characteristic data of the independent variable in the tth year, y t Yes and X t The corresponding dependent variable represents the admission score position value in the province in the tth year, t = (1, 2, ..., T+1), and T represents time.
[0017] Furthermore, the predicted value M' of M and the predicted value Y' of Y are calculated respectively by the ARIMA model, and the calculation process is expressed as follows:
[0018]
[0019] Y'=(y'1,y'2,…,y'′ T ) T
[0020] Among them, X' t =(x' t1 ,x' t2 ,…,x'′ tN ) represents the ARIMA forecast value of the independent variable characteristic data in the tth year, y' t Yes and X' t The corresponding ARIMA forecast value of the dependent variable, t = (1, 2, ..., T+1).
[0021] Furthermore, the error matrix ΔM of M and the error vector ΔY of Y are calculated according to M' and Y', which are respectively expressed as:
[0022]
[0023] ΔY=(y1-y'1,y2-y'2,…,y T -y' T ) T
[0024] Where Δx tj =x tj -x' tj , t=(1,2,...,T+1), j=(1,2,...,N).
[0025] Furthermore, the matrix ΔX composed of the first T rows of data of ΔM is used as the independent variable and ΔY is used as the dependent variable to construct an LSTM residual model, wherein the matrix ΔX is expressed as:
[0026]
[0027] Then the LSTM residual model is expressed as:
[0028] Δy t =f lstm (Δx t , Δx t-1 ,…,ΔX t-p+1 )
[0029] Where ΔX t =(Δx t1 ,Δx t2 ,…,Δx tN ) represents the error between the true value and the predicted value of the independent variable characteristic data in the tth year, Δy t It is the prediction error of the dependent variable obtained by building the LSTM model through ΔX.
[0030] Furthermore, the autoregressive coefficient p is used as the time step of the LSTM model, and the matrix composed of the last p rows of ΔM is input into the LSTM residual model to calculate the error value Δy of the position value of the admission line in year T+1 T+1 , where the matrix consisting of the last p rows of ΔM is expressed as:
[0031]
[0032] Then the error value Δy of the position value of the admission line in year T+1 is obtained T+1 =Δy T+1 =f lstm (x test ).
[0033] Furthermore, the predicted value y' combined with the T+1 year admission line position value T+1 and error value Δy T+1 , obtain the predicted value of the college admission score in T+1 year, which specifically includes the following steps:
[0034] The predicted value of the T+1 year admission line position value y' T+1 and error value Δy T+1 Add together to get the predicted value y of the final admission line position value in year T+1 T+1 , which is represented by y T+1 =y' T+1 +Δy T+1 ;
[0035] Using Min-Max normalization inverse operation on y T+1 Processing is performed to obtain an estimated value of the position value of the filing line;
[0036] According to the one-point-one-section table, the corresponding score of the admission line position value in T+1 year is obtained.
[0037] The present invention also includes a system for predicting the admission score of a college, including:
[0038] A selection module is used to select variables that have affected the college admission score in previous years; the variables include the provincial control line and the total number of college entrance examination candidates as independent variables and the admission score position value as the dependent variable;
[0039] A standardization module is used to standardize the independent variables and dependent variables to obtain an independent variable matrix M and a dependent variable column vector Y with time as rows and feature variables as columns, respectively;
[0040] The first prediction value calculation module is used to input the independent variable matrix M and the dependent variable column vector Y into the ARI MA model to obtain the predicted value M' of M and the predicted value Y' of Y, as well as the predicted value y' of the position value of the admission line in year T+1 T+1 ;
[0041] The LSTM residual model construction module is used to obtain the error matrix ΔM of M and the error vector ΔY of Y based on M and its predicted value M', Y and its predicted value Y', and at the same time obtain the autoregressive coefficient p of the ARI MA model of the dependent variable column vector Y; the matrix ΔX composed of the first T rows of data of ΔM is used as the independent variable and ΔY is used as the dependent variable to construct the LSTM residual model;
[0042] The error value calculation module is used to use the autoregressive coefficient p as the time step of the LSTM model, and input the matrix composed of the last p rows of ΔM into the LSTM residual model to obtain the error value Δy of the position value of the T+1 year enrollment line T+1 ;
[0043] The second prediction value calculation module is used to combine the prediction value y' of the position value of the T+1 year admission line T+1 and error value Δy T+1 , obtain the predicted value of the college admission line in T+1 year.
[0044] The present invention also includes a computer device for predicting the college admission score, including: a memory, a processor, and a computer program stored in the memory, and the processor implements the steps of the college admission score prediction method when executing the computer program.
[0045] The present invention also includes a readable storage medium, which stores a computer program. The computer program includes program instructions. When the program instructions are executed by a processor, they are used to execute the steps of the method for predicting the college admission score.
[0046] The present invention provides a method, system, device and medium for predicting the admission score of a college, which has the following beneficial effects:
[0047] The present invention divides the variables that have affected the college admission score in previous years into independent variables and dependent variables, uses the prediction ability of the ARIMA model for linear time series to extract the linear relationship between the independent variables and the dependent variable time series and makes a prediction, obtains the predicted value of the admission score position, and then predicts the error of the admission score position value according to the LSTM model, combines the prediction ability of the ARIMA model for linear time series and the prediction of the LSTM model for nonlinear time series, thereby completing the complete extraction of time series information and completing the final prediction of the college entrance examination admission score. Compared with the separate ARIMA model and LSTM model, the combined model has higher prediction accuracy. BRIEF DESCRIPTION OF THE DRAWINGS
[0048] Figure 1 Flow chart of the combined model structure in an embodiment of the present invention. DETAILED DESCRIPTION
[0049] The technical solutions in the embodiments of the present invention will be described clearly and completely below in conjunction with the drawings in the embodiments of the present invention. Obviously, the described embodiments are only part of the embodiments of the present invention, rather than all the embodiments.
[0050] The present invention proposes a prediction model for the college admission score based on ARIMA and LSTM models. The model uses the prediction ability of the ARIMA model for linear time series and the accuracy of the LSTM model for nonlinear prediction. First, the ARIMA model is used to extract the linear relationship of the time series and predict it, and then the LSTM model is used to predict the error of the time series, thereby completing the complete extraction of the time series information, completing the final prediction of the college admission score, and improving the prediction accuracy; specifically, the following steps are included:
[0051] Step S1: Determine the variables that affect the college admission line, such as the admission line position value admlinerank, the provincial control line controlline, the total number of college entrance examination candidates totalmount, etc., as shown in Table 1, and vectorize the provincial control line, the total number of college entrance examination candidates, etc. by column to obtain the admission line related variable matrix, and then perform Min-Max standardization on the variable matrix and the admission line position value, and the standardized matrix is expressed as:
[0052]
[0053] The position value of the filing line is expressed as:
[0054] Y=(y1,y2,…,y T ) T
[0055] Among them, X t =(x t1 ,x t2 ,…,x tN ) represents the characteristic data of the independent variable in the tth year, y t Yes and X t The corresponding dependent variable represents the admission score position value in the province that year, t=1, 2,..., T+1.
[0056] Table 1 Description of relevant characteristic variables
[0057]
[0058] Step S2: Establish an ARIMA model for the column vector Y and the matrix M by column, and obtain the ARIMA model expression set F = {f1, f2, ..., f N} and the ARIMA model expression of the dependent variable Y y According to the obtained ARIMA model fi , i=(1,2,…,N,y), recalculate the matrix M to obtain the characteristic matrix M and the predicted value of the dependent variable Y of the ARIMA model:
[0059]
[0060] Y'=(y'1,y'2,…,y' T ) T .
[0061] Step S3: According to the model f y , we can get the model's prediction y for the position value of the admission score in year T+1 T+1 And its autoregression coefficient p. The error value between the matrix M and the dependent variable Y and the prediction matrix M' and the dependent variable predicted value Y' is:
[0062]
[0063] ΔY=(y1-y'1,y2-y'2,…,y T -y' T ) T ,
[0064] Where Δx tj =x tj -x' tj , t=(1,2,...,T+1), j=(1,2,...,N), ΔM and ΔY represent the residual matrix of the independent variable matrix M and the dependent variable Y respectively.
[0065] Step S4: Assume the matrix:
[0066]
[0067] The residual independent variable matrix is t = (1, 2, ..., T), the residual matrix ΔX is taken as the independent variable, ΔY is taken as the dependent variable, and the LSTM model of the time series residual is constructed. The relationship between ΔX and ΔY is:
[0068] Δy t =f lstm (ΔX t , ΔX t-1 ,…,ΔX t-p+1 )
[0069] Step S5: From the relationship between ΔX and ΔY, it can be seen that the dependent variable residual at the current moment is related to the p moments before the current moment. Suppose the test set independent variable matrix is:
[0070]
[0071] Then the residual forecast value of the dependent variable in year T+1 is: Δy T+1 =f lstm (x test ). Therefore, the predicted value of the final admission line position value for the T+1 year is: y T+1 =y' T+1 +Δy T+1 .
[0072] As shown in Table 2, the performance of the ARIMA model, LSTM model and ARIMA-LSTM combination model on the test set is compared.
[0073] Table 2 Model evaluation indicators
[0074]
[0075] Step S6: Use the Min-Max standardization reverse operation to obtain the estimated value of the position value of the filing line: pre_rank=y T+1 ×[max(admlinerank)-min(admlnierank)]+min(admlinerank). Step S7: According to the one-point-one-section table, it is found that there is a corresponding relationship between the admission score and the admission score position value, that is, one position value corresponds to only one score. Therefore, by querying the one-point-one-section table, the score corresponding to the position value can be obtained, and the score is the predicted value pre_admline of the admission score of a certain university in that year.
[0076] In order to verify the effectiveness of the present invention, an experiment was conducted using the admission scores of seven universities including Peking University in Henan Province to verify the performance of the combined model, which specifically includes the following steps:
[0077] Step S1: Determine the variables involved in model training, including the admission line position value, provincial control line, total number of college entrance examination candidates, etc. Determine the known independent variables as the one-point-one-section table, provincial control line and other characteristic information from 2012 to 2023, the known dependent variable is the admission line position value from 2012 to 2022, and what is required is the admission line position value of colleges and universities in 2023. Standardize the data using the Min-Max standardization method to obtain the independent variable matrix M with behavior time and column as characteristic variables and the dependent variable column vector Y.
[0078] Step S2: Establish ARIMA models for column vector Y and matrix M respectively, and use the established models to predict them, and obtain the predicted value M' of matrix M and the predicted value Y' of column vector Y, as well as the predicted value Y' of column vector Y in 2023. 2023 .
[0079] Step S3: Calculate the prediction error of the matrix M and the column vector Y according to the true value and the predicted value, obtain the error matrix ΔM and the error vector ΔY, and at the same time obtain the autoregressive coefficient p of the ARIMA model of the column vector Y, and use p as the time step of the LSTM model.
[0080] Step S4: Using the matrix ΔX composed of the first n-1 rows of data of ΔM as the independent variable and ΔY as the dependent variable, an LSTM model between ΔX and ΔY is established to obtain the model relationship between the two.
[0081] The matrix composed of the last p rows of ΔM is used as the independent variable of the test set, and it is input into the LSTM model obtained in step S4 to obtain the error estimate Δy 2023 The estimated value of the position value of the filing line by the ARIMA model is added to the error estimated value of the position value of the filing line by the LSTM model to obtain the standardized filing line position value y 2023 =Y' 2023 +Δy 2023 .
[0082] Step S6: Perform an inverse normalization operation on it according to the Min-Max normalization method to obtain an estimated value pre_rank of the admission line position value.
[0083] Step S7: According to the score table published in 2023, and based on the correspondence between the admission line position value and the admission line, the score corresponding to the predicted value of the admission line position value is obtained. This score is the admission line pre_admline of a certain university in Henan Province that year.
[0084] In order to more intuitively demonstrate the prediction effect of the model in other provinces, the ARIMA model, LSTM model and the combined model were used for comparative experiments, and the error mean, confidence interval and correlation coefficient were used as evaluation indicators to evaluate the performance of the three models. The results are shown in Table 3.
[0085] Table 3 Evaluation indicators of Henan Province model
[0086]
[0087] Based on the same inventive concept, the present invention also proposes a system for predicting the admission score of a college, including:
[0088] The selection module is used to select the variables that have affected the college admission line in previous years; the variables include the provincial control line and the total number of college entrance examination candidates as independent variables and the admission line position value as the dependent variable.
[0089] The standardization module is used to standardize the independent variables and dependent variables to obtain the independent variable matrix M and the dependent variable column vector Y with behavior time and column as characteristic variables.
[0090] The first prediction value calculation module is used to calculate the prediction value M' of the independent variable matrix M and the prediction value Y' of the dependent variable column vector Y through the ARIMA model, as well as the prediction value Y' of the position value of the admission line in year T+1 T+1 .
[0091] The calculation module is used to calculate the error matrix ΔM and the error vector ΔY of the matrix M and the column vector Y respectively according to M' and Y', and obtain the autoregressive coefficient p of the ARIMA model of the column vector Y.
[0092] The LSTM residual model construction module is used to construct the LSTM residual model with the matrix ΔX composed of the first T rows of data of ΔM as the independent variable and ΔY as the dependent variable.
[0093] The error value calculation module is used to input the matrix composed of the last p rows of ΔM into the LSTM residual model to calculate the error value Δy of the position value of the admission line in year T+1 T+1 .
[0094] The second prediction value calculation module is used to calculate the predicted value Y' of the position value of the T+1 year admission line T+1 and error value ΔyT +1 Add together to get the final predicted value of the admission line position value.
[0095] The prediction module is used to predict the admission score of colleges and universities in a certain future year based on the final predicted value of the admission score position value.
[0096] Based on the same inventive concept, the present invention also proposes a computer device for predicting the college admission score, including: a memory, a processor, and a computer program stored in the memory. When the processor executes the computer program, the steps of the college admission score prediction method are implemented.
[0097] Based on the same inventive concept, the present invention also proposes a readable storage medium, which stores a computer program. The computer program includes program instructions. When the program instructions are executed by a processor, they are used to execute the steps of the method for predicting the college admission score.
[0098] The above description is only a preferred specific implementation manner of the present invention, but the protection scope of the present invention is not limited thereto. Any technician familiar with the technical field can make equivalent replacements or changes according to the technical scheme and inventive concept of the present invention within the technical scope disclosed by the present invention, which should be covered by the protection scope of the present invention.
Claims
1. A method for predicting the admission score of a college, characterized in that: The following steps are involved: Select the variables that have affected the college admission score in previous years; the variables include the provincial control line and the total number of college entrance examination candidates as independent variables and the admission score position value as the dependent variable; The independent variables and dependent variables are standardized to obtain an independent variable matrix M and a dependent variable column vector Y with time as rows and characteristic variables as columns, respectively; Input the independent variable matrix M and the dependent variable column vector Y into the ARIMA model to obtain the predicted value M' of M and the predicted value Y' of Y, as well as the predicted value y' of the position value of the admission line in year T+1 T+1 ; According to M and its predicted value M', Y and its predicted value Y', we get the error matrix ΔM of M and the error vector ΔY of Y, and at the same time, we get the autoregressive coefficient p of the ARIMA model of the dependent variable column vector Y; we use the matrix ΔX composed of the first T rows of data of ΔM as the independent variable and ΔY as the dependent variable to build the LSTM residual model; The autoregressive coefficient p is used as the time step of the LSTM model, and the matrix composed of the last p rows of ΔM is input into the LSTM residual model to obtain the error value Δy of the position value of the admission line in year T+1. T+1 ; Combined with the predicted value y' of the T+1 year admission line position value T+1 and error value Δy T+1 , obtain the predicted value of the college admission line in T+1 year.
2. A method for predicting the admission score of a college according to claim 1, characterized in that: The independent variables and dependent variables are standardized by using the Min-Max standardization method to obtain the independent variable matrix M and the dependent variable column vector Y with behavior time and column as characteristic variables, which are respectively expressed as: Y=(y1,y2,…,y T ) T Among them, X t =(x t1 ,x t2 ,…,x tN ) represents the characteristic data of the independent variable in the tth year; y t Yes and X t The corresponding dependent variable represents the admission score position value in the province in the tth year; t = (1, 2, ..., T+1); T represents time.
3. A method for predicting the admission score of a college according to claim 2, characterized in that: The independent variable matrix M and the dependent variable column vector Y are input into the ARIMA model to obtain the predicted value M' of M and the predicted value Y' of Y, specifically including: Y'=(y'1,y'2,…,y' T ) T Among them, X' t =(x' t1 ,x' t2 ,…,x' tN ) represents the ARIMA forecast value of the independent variable characteristic data in the tth year; y' t Yes and X' t The corresponding ARIMA forecast value of the dependent variable; t = (1, 2, ..., T+1).
4. A method for predicting the admission score of a college according to claim 3, characterized in that: The error matrix ΔM of M and the error vector ΔY of Y are obtained according to M and its predicted value M', Y and its predicted value Y', which are respectively expressed as: ΔY=(y1-y'1,y2-y'2,…,y T -y' T ) T Where Δx tj =x tj -x' tj ;t=(1,2,...,T+1); j=(1,2,...,N).
5. A method for predicting the admission score of a college according to claim 4, characterized in that: The matrix ΔX composed of the first T rows of data of ΔM is used as the independent variable and ΔY is used as the dependent variable to construct an LSTM residual model, wherein the matrix ΔX is expressed as: Then the LSTM residual model is expressed as: Δy t =f lstm (ΔX t ,ΔX t-1 ,…,ΔX t-p+1 ) Where ΔX t =(Δx t1 ,Δx t2 ,…,Δx tN ) represents the error between the true value and the predicted value of the independent variable characteristic data in the tth year, Δy t It is the prediction error of the dependent variable obtained by building the LSTM model through ΔX.
6. A method for predicting the admission score of a college according to claim 5, characterized in that: The autoregressive coefficient p is used as the time step of the LSTM model, and the matrix composed of the last p rows of ΔM is input into the LSTM residual model to obtain the error value Δy of the position value of the admission line in year T+1 T+1 , where the matrix consisting of the last p rows of ΔM is expressed as: Then the error value Δy of the position value of the admission line in year T+1 is obtained T+1 =Δy T+1 =f lstm (x test ).
7. The method for predicting the admission score of a college according to claim 1, characterized in that: The predicted value y' combined with the T+1 year admission line position value T+1 and error value Δy T+1 , obtain the predicted value of the college admission score in T+1 year, which specifically includes the following steps: The predicted value of the T+1 year admission line position value y' T+1 and error value Δy T+1 Add together to get the predicted value y of the final admission line position value in year T+1 T+1 , which is represented by y T+1 =y' T+1 +Δy T+1 ; Using Min-Max normalization inverse operation on y T+1 Processing is performed to obtain an estimated value of the position value of the filing line; According to the one-point-one-section table, the corresponding score of the admission line position value in T+1 year is obtained.
8. A system for predicting the admission score of a college, characterized in that: include: A selection module is used to select variables that have affected the college admission score in previous years; the variables include the provincial control line and the total number of college entrance examination candidates as independent variables and the admission score position value as the dependent variable; A standardization module is used to standardize the independent variables and dependent variables to obtain an independent variable matrix M and a dependent variable column vector Y with time as rows and feature variables as columns, respectively; The first prediction value calculation module is used to input the independent variable matrix M and the dependent variable column vector Y into the ARI MA model to obtain the predicted value M' of M and the predicted value Y' of Y, as well as the predicted value y' of the position value of the admission line in year T+1 T+1 ; The LSTM residual model construction module is used to obtain the error matrix ΔM of M and the error vector ΔY of Y based on M and its predicted value M', Y and its predicted value Y', and at the same time obtain the autoregressive coefficient p of the ARI MA model of the dependent variable column vector Y; the matrix ΔX composed of the first T rows of data of ΔM is used as the independent variable and ΔY is used as the dependent variable to construct the LSTM residual model; The error value calculation module is used to use the autoregressive coefficient p as the time step of the LSTM model, and input the matrix composed of the last p rows of ΔM into the LSTM residual model to obtain the error value Δy of the position value of the T+1 year enrollment line T+1 ; The second prediction value calculation module is used to combine the prediction value y' of the position value of the T+1 year admission line T+1 and error value Δy T+1 , obtain the predicted value of the college admission line in T+1 year.
9. A computer device for predicting the admission score of a college, characterized in that: include: A memory, a processor and a computer program stored in the memory, wherein the processor implements the steps of the method for predicting the college admission score as described in any one of claims 1 to 7 when executing the computer program.
10. A readable storage medium, characterized in that: The readable storage medium stores a computer program, which includes program instructions. When the program instructions are executed by a processor, they are used to execute the steps of the method for predicting the college admission score as described in any one of claims 1-7.