Yazarlar: Denizhan DEMİRKOL, Elif KARTAL, Çağla ŞENELER, Sevinç GÜLSEÇEN
Anahtar Kelimeler:supervised learning, usability, machine learning, student information systems, system usability scale
Özet: System usability is one of the key elements that should be focused on, especially during the design and test phases of a system, because it provides feedback to system administrators in order to improve the system. In the literature, System Usability Scale (SUS) is widely used as the gold standard method to evaluate system usability. Today, machine learning, which is one of the subfields of artificial intelligence, also provide new perspectives on the evaluation of system usability. In this study, it is aimed to predict usability of a Student Information System (SIS) by using machine learning techniques. In the study method, the Cross-Industry Standard Process for Data Mining (CRISP-DM) steps have been followed. Analysis are performed on two different dataset namely “sus1” and “sus0”. “sus1” dataset is consisted of demographic characteristics (age, gender, department) of 324 students using a SIS of a foundation university in Turkey, also their responses to the Turkish version of the SUS. “sus0” includes only responses to the Turkish version of the SUS. C4.5 Decision Tree Algorithm, Naive Bayes Classifier and k-Nearest Neighbor Algorithm are used to create models and their performance are evaluated. The best performance was obtained on the “sus0” data set with 80% to 20% hold-out method (accuracy = 0.698, F-measure = 0.796 for k = 20).
Dergi editörleri editör girişini kullanarak sisteme giriş yapabilirler. Editör girişi için tıklayınız.