ANALISSIS SENTIMEN KEPUASAN PELANGGAN ONLINE SHOP MENGGUNAKAN ALGORITMA NAÏVE BAYES CLASSIFIER DAN K-NEAREST NEIGHBOR (Studi Kasus : Twitter Resmi Shopeecare)

Inaratul Mahbubah, NIM.: 16650069 (2020) ANALISSIS SENTIMEN KEPUASAN PELANGGAN ONLINE SHOP MENGGUNAKAN ALGORITMA NAÏVE BAYES CLASSIFIER DAN K-NEAREST NEIGHBOR (Studi Kasus : Twitter Resmi Shopeecare). Skripsi thesis, UIN SUNAN KALIJAGA YOGYAKARTA.

Preview	Text (ANALISSIS SENTIMEN KEPUASAN PELANGGAN ONLINE SHOP MENGGUNAKAN ALGORITMA NAÏVE BAYES CLASSIFIER DAN K-NEAREST NEIGHBOR (Studi Kasus : Twitter Resmi Shopeecare)) 16650069_BAB-I_IV-atau-V_DAFTAR-PUSTAKA.pdf Download (2MB) | Preview
	Text (ANALISSIS SENTIMEN KEPUASAN PELANGGAN ONLINE SHOP MENGGUNAKAN ALGORITMA NAÏVE BAYES CLASSIFIER DAN K-NEAREST NEIGHBOR (Studi Kasus : Twitter Resmi Shopeecare)) 16650069_BAB-II_sampai_SEBELUM-BAB-TERAKHIR.pdf Restricted to Registered users only Download (3MB)

Abstract

Shopee is one of the mostly used online marketplace in Indonesia. People around the world are most likely to buy products through online marketplace nowadays, including Indonesians. It happens because the citizens find online shopping is easier and more practical. Other than that, some people think the products are cheaper and the availability of many discounted prices given by the online stores. In this research, sentiment analysis was carried out using 1015 tweets as training data and 480 tweets as testing data. The data was tested using the Naïve Byes Classifier method and K-NN using the confusion matrix. Data tested using the Naïve Bayes Classifier using confusion matrix obtains an accuracy rate of 86.27% with a split ratio of 90:10, an accuracy rate of 88.17% with a split ratio of 80:20, and an accuracy rate of 85.90% with a split ratio of 70:30. Whilst the data processed using K-NN using a configuration matrix obtains an accuracy rate of 84.31% with a split ratio of 90:10, an accuracy rate of 87.68% with a split ratio of 80:20, and an accuracy rate of 87.21% with a ratio split 70:30. The results of the implementation of the 480 new data tweets obtained on the Naïve Bayes Classifier are 25.6% positive data and 74.4% negative data. Meanwhile, the implementation of K-NN are 40.2% positive data and 59.8% negative data.

Share this knowledge with your friends :

Actions (login required)

View Item