Klasifikasi Emosi Pelajar terhadap Metode Pembelajaran Online atau Daring Menggunakan K-Nearest Neighbor dan Mutual Information

Abstract

The COVID-19 or Coronavirus disease 2019 pandemic has spread to almost all parts of the world. COVID-19 is a new type of corona virus called SARS-CoV-2. Since early January, it is likely that COVID-19 has entered Indonesia. COVID19 is thought to be spread through the air. One of the recommended steps to prevent getting COVID-19 is to keep your distance. This caused the government to issue a decision for people to carry out activities from home. This allows students and teachers to carry out learning activities from home using online or online methods. Learning activities carried out with this method raise many pros and cons for students. On social media such as Instagram, Twitter, and Facebook, many arguments have emerged regarding this learning method. The number of these arguments can be used by teaching staff to carry out evaluations so that in the future they can be even better. However, it would be time-consuming to summarize these arguments if read one by one. Knowing the learner's emotions from his argument will make the evaluation more effective. Emotion recognition can be done by classification. Classification is a grouping based on the same characteristics of objects, ideas, books or objects into a certain class. Text mining is one of the methods used in the classification process. In this study, K-Nearest Neighbor is used as a method for text mining. The whole scenario of the method used is preprocessing. Then weighting is carried out using Mutual Information and K-Nearest Neighbor to carry out the emotion classification process. This study resulted in an accuracy rate of 75.49% with a total of 2040 datasets. Model accuracy was validated with k-fold cross validation. With a k-fold value, k=10 produces an accuracy of 76