Personalized College Entrance Examination Filling Recommendation Based On Machine Learning

Filling in the college entrance examination is an indispensable part on the long march of the college entrance examination,and the quality of the application determines the future development direction of candidates.Therefore.It’s a question of concern for all candidates and parents on how to search for useful information in the massive data of college entrance examination.By analyzing the historical admission data and the personalized requirements of candidates and parents for major selection,this paper excavates effective information from massive faculties and professional data.and uses machine learning algorithms to design personalized filling recommendation methods.so as to provide practical help for candidates and parents in the process of filling in the college entrance examination.Firstly.this paper obtains and integrates the admission score data of each institution and major from school official website and related websites.and then selects appropriate measures to generate time series data.When predicting the score line.the ARIMA algorithm is graded and fitted,and the fitted residual sequence is used as the input of the machine learning algorithm for further regression to fit the nonlinear part.The results of the two models are synthesized to generate the final prediction result.Different algorithms are tried to obtain the optimal combination model:ARIMA+XGBoost,which is used as the prediction model to obtain the prediction result in 2023.Secondly.to solve the problem of the poor understanding of majors and the colloquial requirements of major selection by candidates and parents.first of all.TF-IDF algorithm is used to obtain text keywords.and then keyword search and word segmentation technology are used to locate majors.last but not least.in order to expand the scope of major recommendations.BERT model is used for vector representation and cosine similarity is calculated to retrieve similar majors.Finally.according to the above research methods.a visual recommendation webpage is generated,merging two parts of the research content,and the practicability and effectiveness of the study are verified through page display and recommendation case analysis.