Open AccessOptimasi Seleksi Fitur dengan Teknik Reduksi Dimensi pada Klasifikasi Abstrak Jurnal
Author(s) -
Syukriyanto Latif
Publication year - 2019
Publication title -
jurnal penelitian enjiniring fakultas teknik unhas
Language(s) - English
Resource type - Journals
eISSN - 2685-4104
pISSN - 1411-6243
DOI - 10.25042/jpe.052018.08
Subject(s) - computer science , weighting , dimensionality reduction , thresholding , computation , feature selection , artificial intelligence , pattern recognition (psychology) , dimension (graph theory) , naive bayes classifier , data mining , algorithm , mathematics , physics , combinatorics , acoustics , image (mathematics) , support vector machine
The purpose of this research is to know dimension reduction parameter value at feature selection so as to improve accuracy and reduce computation time. This system uses text mining technology that extracts text data to find information from a set of documents. Word weighting and Term Reduction Technique The term Frequency Thresholding is used in the feature selection process, while in the classification process using the Naive Bayes algorithm. the abstract of the journal is categorized into 3 namely Data Mining (DM), Intelligent Transport System (ITS) and Multimedia (MM). The total number of test data and training data is 150 data. The best classification results are obtained when the dimension reduction parameter value is 30%. At that condition obtained an average accuracy of 87.33% with a computation time of 4 minutes 12 seconds.