Penerapan Algoritma Naïve Bayes Untuk Melakukan Analisis Sentimen Pada PT Pos Indonesia (Persero)

Abstract

Pos Indonesia is the oldest shipping service that is widely known to the public, so it has different opinions on the performance of postal expeditions. The author identifies the problem in this research, namely the public's sentiment on the services of PT. Pos Indonesia (Persero) which is found on the Twitter social media platform and the number of positive or negative sentiments towards the services of PT. Pos Indonesia (Persero). Researchers use social media Twitter as a medium to get data to examine the performance of the Indonesian Post. In this research, the writer intends to analyze the sentiment towards PT. POS Indonesia (Persero) as an identification material for negative and positive opinions by using the nave Bayes algorithm to determine the service performance of PT. POS Indonesia (Persero). Researchers also use CRISP-DM as a data processing method and use rapid miner applications to obtain, process and produce positive and negative classifications. Classification of data in this study took 141 tweets discussing PT. POS Indonesia (Persero) on Twitter media by using the keyword Pos Indonesia. The results of this study resulted in a positive sentiment value of 63% and a negative 37%. With the highest accuracy, the 80:20 data split method uses the Naive Bayes algorithm of 64.29%.