IMPLEMENTATION OF INFORMATION GAIN AND PARTICLE SWARM OPTIMIZATION UPON COVID-19 HANDLING SENTIMENT ANALYSIS BY USING K-NEAREST NEIGHBOR

Abstract

In early 2020, a new virus from Wuhan, China, identified as the coronavirus or COVID-19, shocked the entire world. (Coronavirus Disease 2019). The government has made various attempts to combat this outbreak, despite the fact that the government's involvement in combating Covid-19 has many benefits and disadvantages. One of the most commonly debated subjects on Twitter is the Indonesian government's response to the Covid-19 virus. This research compares the k-nearest neighbor classification technique, Information Gain feature selection with the K-Nearest Neighbor classification algorithm, and Information Gain feature selection and Particle Swarm Optimization optimization with the K-Nearest Neighbor classification algorithm. Comparisons are performed to determine which method is more accurate. Because it is frequently used for text and data categorization, the K-Nearest Neighbor algorithm was selected. The K-Nearest Neighbor algorithm has flaws, including the ability to be fooled by irrelevant characteristics and being less than ideal in finding the value of k. The selection of the Information Gain feature could indeed solve this issue by decreasing Terms that are less important and to optimize the K-Nearest Neighbor categorization, an optimization method with the Particle Swarm Optimization algorithm is employed to maximize the K-Nearest Neighbor classification. According to the results of this research, the K-Nearest Neighbor categorization with Information Gain feature selection and Particle Swarm Optimization optimization is better than the K-Nearest Neighbor model without selecting features and without optimization and is better than the K-Nearest Neighbor model with Information Gain selecting features, notably 87,33% with a value of K 5.