Evaluation of the last grade and attempts in online-learning environment

The purpose of this thesis was to develop models and methods, which evaluated learning in an online environment. In order to reach the research goal, data from tutor-web.net, an online repository containing online lectures and assignments, with questions for various subjects in mathematics, was obta...

Full description

Bibliographic Details
Main Author: Duc Hung Bui 1990-
Other Authors: Háskóli Íslands
Format: Thesis
Language:English
Published: 2020
Subjects:
Online Access:http://hdl.handle.net/1946/40389
Description
Summary:The purpose of this thesis was to develop models and methods, which evaluated learning in an online environment. In order to reach the research goal, data from tutor-web.net, an online repository containing online lectures and assignments, with questions for various subjects in mathematics, was obtained and investigated. The data includes results for students enrolled in the course Applied Mathematical Analysis at the University of Iceland in 2017 and 2018. First, descriptive statistics were used to determine which factors influenced students’ performance. Next, models were developed to predict the final grades and the number of attempts. The result shows that tuning the system to a faster increase in difficulty leads to learning for a longer time (more attempts), as well as a grade which is likely more indicative of actual knowledge as opposed to when students receive easy questions for too long. Grades in the tutor-web are based on a weighted combination of recent answers and increasing the number of answers and high weight to the most recent questions in the grade leads students to answer more questions. Markmið þessarar ritgerðar var að þróa líkön og aðferðir sem meta lærdóm í kennslukerfi á vef. Notast var við gögn frá tutor-web.net, sem geymir kennsluefni og æfingar í ýmsum greinum, meðal annars stærðfræði, sem var notað í ritgerðinni. Gögnin innihalda svör nemenda við Háskóla Íslands, sem voru skráðir í áfangann Hagnýtt stærðfræðigreining árin 2018 og 2019. Fyrst var lýsandi tölfræði notuð til að kanna, hvaða þættir hefðu áhrif á frammistöðu nemenda. Síðan voru tölfræðileg líkön sett upp til að spá fyrir um lokaeinkunn, fjölda tilrauna og meta tilsvarandi stika. Ef kerfið er stillt á að þyngja spurningar hratt þurfa nemendur að læra lengur (fleiri tilraunir) og fá lægri lokaeinkunn, sem er líklegast betri vísbending um raunverulega kunnáttu heldur en ef nemendur fá of léttar spurningar of lengi. Einkunnakerfið byggir á vegnu meðaltali síðustu svara og í ljós kemur að nemendur svara fleiri spurningum ...