{"id":7,"date":"2019-06-24T08:44:11","date_gmt":"2019-06-24T05:44:11","guid":{"rendered":"https:\/\/lapsus.kapsi.fi\/blog\/?p=7"},"modified":"2020-05-04T13:20:44","modified_gmt":"2020-05-04T10:20:44","slug":"machine-learning-to-predict-student-graduation-in-finland-news-from-aalto-university","status":"publish","type":"post","link":"https:\/\/lauriviitanen.kapsi.fi\/blog\/machine-learning-to-predict-student-graduation-in-finland-news-from-aalto-university\/","title":{"rendered":"Machine Learning to Predict Student Graduation in Finland: News from Aalto University"},"content":{"rendered":"\n
\"\"<\/figure>\n\n\n\n

Machine learning can predict students\u2019 graduation with 74% probability<\/h2>\n\n\n\n

8.6.2016<\/p>\n\n\n\n

In his Master\u2019s thesis, Lauri Viitanen analysed information about over 30,000 students from Metropolia to predict their graduation dates.<\/strong><\/p>\n\n\n\n

In his thesis, Lauri Viitanen<\/strong>, a student of Computer Science and Engineering, applied survival models<\/a>\n to the information in the student register at Metropolia University of \nApplied Sciences with the aim of finding variables that can best \ndistinguish the students who will graduate from those who will drop out.\n In addition, he also aimed to find variables that predict the remaining\n study time the most accurately.<\/p>\n\n\n\n

In his thesis, Lauri Viitanen<\/strong>, a student of Computer Science and Engineering, applied survival models<\/a> to the information in the student register at Metropolia University of Applied Sciences with the aim of finding variables that can best distinguish the students who will graduate from those who will drop out. In addition, he also aimed to find variables that predict the remaining study time the most accurately.<\/p>\n\n\n\n

“In my thesis, I composed a model that will predict \nthe graduation date of students at Metropolia University of Applied \nSciences based on their study performance in their first year and other \nexplanatory factors. The explanatory factors in the model included, for \nexample, age, gender, the field of a prior study right, whether or not \nthe studies begun during the spring term, credit points accumulated \nduring the first year and the weighted average of grades.”<\/p>\n\n\n\n

In the thesis, students were classified using naive \nBayes classifiers, generalised linear models, support vector machine \nclassifiers and Gaussian processes. Gaussian processes have not been \napplied to a similar material before. Although survival models are very \nwell suited to this kind of longitudinal studies, Gaussian processes can\n be used to increase their flexibility. It would be useful to examine \nhow much exactly would accuracy be improved by the increased \nnon-linearity of the model.<\/p>\n\n\n\n

“In my thesis, I also compared how accurately these \nfour machine learning methods could classify students right after their \nfirst year of studies either to those that will eventually graduate or \nthose that will drop out.”<\/p>\n\n\n\n

An accuracy of 74 per cent could be achieved with the \nbest machine learning method, which was the support vector machine. In \nother words, for three out of four students, the end of their study \nright could be estimated correctly as early as after the first year of \nstudies. Completing extra credits during the first year increased the \ngraduation probability more than an improvement of the grade average by \none grade. It has not been studied before what kind of influence the \ntotal number of credits accumulated has on graduation in comparison to \nother factors.<\/p>\n\n\n\n

“Metropolia University of Applied Sciences intends to \nutilise the results of the thesis in planning their budget, as \ninaccurate estimation of graduation dates makes it more difficult to \npredict future funding. By using the student-specific model, Metropolia \nintends to reduce the error. The results are also likely to be utilised \nin the students\u2019 workspaces so that it will be easy for \nstudy counsellors and group leaders to detect the students whose \nprogress they should pay most attention to,” Viitanen estimates the \npossibilities provided by the study.<\/p>\n\n\n\n

The results of the study are in line with earlier \nstudies: the grade average and the student\u2019s gender are significant \nvariables when estimating graduation probability. Girls are more likely \nto graduate and they graduate faster than boys, regardless of the \nsubject. An increase in the student\u2019s age had a negative effect on \ngraduation probabilities both in this study and in almost all earlier \nstudies.<\/p>\n\n\n\n

References<\/p>\n\n\n\n