Analisis Sentimen Kinerja KPU Pemilu 2019 Menggunakan Algoritma K-Means Dengan Algoritma Confix Stripping Stemmer

Abstract

The performance of the KPU in the 2019 elections was a lively conversation between the general public and the political elite. Many people or parties commented on the process of calculating the results of the 2019 elections through social media. KPU's Facebook fan page is also busy being attacked by positive or negative comments. Sentiment analysis of public opinion can be investigated to find out how much percentage of a positive and negative sentiment of this policy through comments that have been sent to social-social media. Data analyzed were 200 data taken from Facebook KPU, divided into 150 training data and 50 test data. This study uses the K-Means algorithm with a value of k = 2 to determine the final sentiments of positive and negative, the Levenshtein Distance algorithm for word normalization and the Confix Stripping Stemmer algorithm in the stemming process. The results obtained from the public sentiment on the performance of the KPU are more negative than positive. The results of the accuracy obtained from the use of the K-Means algorithm are 84% with a lower accuracy value compared to the combination of the algorithm above, namely 86%. Suggestions for further research should use even more data and use the k-fold cross-validation accuracy calculation technique as a further trial

Keywords

• confix stripping stemmer
• facebook
• kpu
• k-means
• levensthein distance
• pemilu
• sentimen analysis.