Optimizing the certainty factor on k-nearest neighbor to determine the learning model during the pandemic

Abstract

In the pandemic era, the lecture must use an online class, not an offline class. The online course has a significant problem in learning because students must adapt to model learning their lesson, students have their education, and teachers have their style. So it will happen throughout the pandemic era. It is a big problem in learning. At the end of the semester, the student usually gives feedback for their lecture on what kind of model they want. The data of feedback is put in a database online learning. The data can be used to know what students need and what course must choose a model for online learning. This research is applied to determine the kind of model learning suitable for students in the pandemic era. The determination process uses a k-nn algorithm that classifies all parameters and data and makes decisions based on all parameters and data. The data in this research is from database learning Universitas Negeri Malang and data feedback from students. All data is processed and calculated by k-nn. This application has a maximum accuracy of 88.37% in testing the variation of the k value, with \mathrmk=3,\ \mathrmk=5, and \mathrmk=8.