2019年世界人口綜述指出台灣出生率在全球200個國家中敬陪末座,2020年國發會公布的「2020年至2070年人口推估」報告,台灣總人口將於2020正式轉呈負成長,出生率下降的問題讓許多私立校院招生缺額大幅增加。在面臨招收不到學生的生存危機下,各校為招生投入非常多心力,加強教學及服務品質,讓學生有良好的學習環境,但休退學生數卻依然持續增加。 另一方面,由轉學生結構模型發現,轉學行為並非單一因素造成,學生在求學階段遭遇的各種問題,促成了學生的轉學。就學生而言,轉學後又必須重新適應新的環境,使學生同時面對課業與環境適應雙重挑戰。對學校而言,一位學生的轉學行為,會降低班上的凝聚力,甚至引起班上同學轉學的連鎖反應,加速了學生的流失。 本研究以某私立校院103至108年學生資料17,619筆,運用人工智慧神經網路多層感知器(Multilayer Perceptron)及支持向量機(Support Vector Machine)兩種監督式學習的模型預測學生在求學期間是否會因故離校的可能性,讓學校能掌握這些學生進行輔導,避免因轉學及休退學而遭遇各種問題,讓學生能安心就學,營造一個良善的學習環境,讓學生能學以致用,迎接新世代的挑戰,同時減少轉學的學生數,達到顧生目的,更進一步吸引更多年輕學子的加入,創造三贏的局面。
Taiwan had the lowest fertility rate of the global 200 countries, according to a 2019 report by the World Population Review, National Development Council released its demographic projections for 2020-2027 that Taiwan will see negative population growth in 2020. Taiwan’s private technological and vocational colleges or universities will suffer new student enrollment decline from low birth rate, and are feared will be shuttered by insufficient number of students, therefore, they endeavor to offer a wide variety of academic resources and services, providing students with excellent educational environment, but the students who are suspended or dropped out are still increasing. In addition, to explore the reason why students transfer to other college is not single factor to be determined, founded by transfer student structural model. The students would transfer to other school as the result of various problems happened in their school life. From the student view, in order to transition well, he must adjust to a new environment, ready to take on the dual challenges of academic curriculum and surroundings. From the school view, the act on student transfer will have a knock-on effect on classmate care, scattering the class unity, further speeding up the student loss. This research is to operate an Artificial Intelligence Neural Network Multilayer Perceptron (MLP) model and Support Vector Machine (SVM) model to predict the student transfer probability in school life according to 17,619 records of 2014-2019 students’ data from one private university, so that the school can use it for counseling solutions to the transfer, suspend, or drop-out problem, enable more students to feel comfortable in school, create a friendly learning atmosphere, instruct students to practice what they have learned, furthermore, in order to face new challenges, the school will achieve a win-win situation to reduce transfer student number, to comfort student retention learning well, and to attract new students.