Analisis sentimen komentar mahasiswa terhadap kinerja dosen menggunakan metode multinomial naïve bayes dan bernoulli naïve bayes
Total View This Week0
Total View This Week0
Institusion
Institut Teknologi Telkom Purwokerto
Author
Fajar, Prayogi Sofiandika
Subject
T Technology (General)
Datestamp
2021-04-21 06:35:59
Abstract :
According to Permenristekdikti NO 44 of 2015 concerning the National Standards of Higher Education which contains conducting periodic monitoring and evaluation activities in order to maintain and improve the quality of the learning process, In IT Telkom Purwokerto applies the assessment of lecturers to the Igracias system, the assessment is done in 2 ways namely scale 1 up to 5 and also comments, therefore the data used in this study are students' comments on the teaching performance of lecturers, which will be labeled positive and negative. The data used are 2177 data that will be divided into 50% of training data and 50% of test data with a comparison of 80% of training data and 20% of training data, the algorithm used in this study is the Multinomial Naïve Bayes and Bernoulli Naïve Bayes where the algorithm used adopts the concept probability and statistics. The final result of this study is to compare the accuracy results of the two algorithms used, and after the data normalization processes are carried out to produce accuracy, it has been obtained that the Multinomial Naïve Bayes 89% and Bernoulli Naïve Bayes 84%. In this case the Multinomial Naïve Bayes algorithm has better performance than Bernoulli Naïve Bayes because the word frequency calculation on the Multinomial Naïve Bayes is counted every word in the document while in the Bernoulli Naïve Bayes the boolean value is calculated to be 1 to exist and value 0 to none. at the final count after all probabilities have been calculated, it will be multiplied by the prior class in the Multinomial Naïve Bayes only the test data is multiplied by the class prior but in the Bernoulli Naïve Bayes all times the training data, the class prior, the training data, this causes differences in accuracy obtained by both methods. From the results obtained it can be concluded that Multinomial Naïve Bayes is better than Bernoulli Naïve Bayes. Keywords:sentiment analysis, student questionnaire, bernoulli naïve bayes, multinomial naïve bayes
According to Permenristekdikti NO 44 of 2015 concerning the National Standards of Higher Education which contains conducting periodic monitoring and evaluation activities in order to maintain and improve the quality of the learning process, In IT Telkom Purwokerto applies the assessment of lecturers to the Igracias system, the assessment is done in 2 ways namely scale 1 up to 5 and also comments, therefore the data used in this study are students' comments on the teaching performance of lecturers, which will be labeled positive and negative. The data used are 2177 data that will be divided into 50% of training data and 50% of test data with a comparison of 80% of training data and 20% of training data, the algorithm used in this study is the Multinomial Naïve Bayes and Bernoulli Naïve Bayes where the algorithm used adopts the concept probability and statistics. The final result of this study is to compare the accuracy results of the two algorithms used, and after the data normalization processes are carried out to produce accuracy, it has been obtained that the Multinomial Naïve Bayes 89% and Bernoulli Naïve Bayes 84%. In this case the Multinomial Naïve Bayes algorithm has better performance than Bernoulli Naïve Bayes because the word frequency calculation on the Multinomial Naïve Bayes is counted every word in the document while in the Bernoulli Naïve Bayes the boolean value is calculated to be 1 to exist and value 0 to none. at the final count after all probabilities have been calculated, it will be multiplied by the prior class in the Multinomial Naïve Bayes only the test data is multiplied by the class prior but in the Bernoulli Naïve Bayes all times the training data, the class prior, the training data, this causes differences in accuracy obtained by both methods. From the results obtained it can be concluded that Multinomial Naïve Bayes is better than Bernoulli Naïve Bayes. Keywords:sentiment analysis, student questionnaire, bernoulli naïve bayes, multinomial naïve bayes
Download
book
BibTex
Latex, Jabref
cloud_download
Original Resource
url resource Institution
