Machine Learning Based Methods Used for Improving Scholar Performance
No Thumbnail Available
In their 1968 study "The Teaching-Learning Paradox: A Comparative Analysis of College Teaching Methods", Robert Dubin and Thomas Taveggia found no evidence to indicate any basis for preferring one teaching method over another as measured by the performance of students on course examinations. The conclusion is based on systematic reanalysis of the data of almost 100 comparative studies of different college teaching methods. The teaching-learning process is virtually a black box in which the teaching methods do not influence scholar performance. So rather than focusing on teaching methods, we propose a method for improving scholar performance by a continuous and intelligent monitoring and assessing process of daily knowledge gains. Recent developments in machine learning and data analysis, allow us the use of techniques for unveiling the strengths and weaknesses in the learning process. Our proposed solution is “micro-assessing” each student and after each course, by sending a 5 minutes-long mini test on their smartphones, collects the result and send it for analysis to a dedicated platform. When the system has sufficient data, it can personalize the tests for each student with focus on areas the student is lacking. It can make recommendations, to teachers, students, school and competent authorities at local or national level. The solution is not a replacement for classical examinations, but it augments the learning experience through interactive and personalized quizzes. The teacher also has a better view over student’s knowledge, thus he can do better assessments overall. Student abnormal deviations could be detected much faster, while competent authorities could assess the impact of their decisions in near-realtime.
scholar performance, assessment model, machine learning, knowledge gain, RLO, IAMA