PERHITUNGAN ANALISIS SENTIMEN BERBASIS KOMPARASI ALGORITMA NAIVE BAYES DAN K-NEAREST NEIGHBOUR BERBASIS PARTICLE SWARM OPTIMIZATION PADA KOMENTAR INSIDEN PEMBALAP MOTOGP 2015

Sari

Media to get information about news MotoGP rider very much like media TV, radio, newspapers, magazines, websites and others. But from most of the media is a media website which is very flexible because it can be accessed from a wide variety of places connected to the Internet, the information provided is up to date and everyone can comment on articles related. The information spreads very fast and is accompanied by the freedom of speech can cause various types of opinions, either negative or positive opinion. Classification techniques of some of the most frequently used is Naive Bayes and k-Nearest Neighbour KNN). Naive Bayes classifier is a simple applying Bayes Theorem to independence (independent) high. K-Nearest Neighbor (KNN) classification algorithm predicts the category of the test sample in accordance with the training sample K nearest neighbor to the test sample, and a judge for the category that has the largest category of probability. Therefore, in this study using the merging feature selection methods, namely particle Swarm Optimization in order to improve the accuracy on Naive Bayes and k-Nearest Neighbour. As for the resulting accuracy Naive Bayes algorithm based on Particle Swarm Optimmization with an accuracy of 82.67%. and k-Nearest Neighbour-based Particle Swarm Optimmization with an accuracy of 71.33% It can be concluded that the application of optimization can improve accuracy. Model in Naive Bayes-based Particle Swarm Optimization can provide solutions to the problems of classification review of public opinion news MotoGP racer incident in order to more accurately and optimally. for the model-based k-Nearest Neighbour Particle Swarm Optimization accuracy decreases.

Keywords : Media, Classification, Naive Bayes, k-Nearest Neighbour, Particle Swarm Optimization, Text Mining.