OPTIMASI ALGORITMA NAÏVE BAYES DENGAN INFORMATION GAIN RATIO UNTUK MENANGANI DATASET BERDIMENSI TINGGI

Abstract

The development of computer science now allows the recording of all business processes in all fields with large storage media. Data in the fields of atronomy, health, economy, government and so on is widely recorded and is increasing from year to year. Data mining is a science that can process data into a representation of knowledge using several mathematical methods or algorithms. One of the main functions of data mining is classification. In the process of classification all old data is used as learning data to infer new data that is not yet fully known. Data which previously has no meaning can become a new knowledge by using data mining classification. Many algorithms can be used in the classification process. One of the algorithms that is proven to be good for the process of classifying high-dimensional data is Naïve Bayes.
In high-dimensional data the many data attributes can affect the results of
classification. The number of relevant data attributes can improve algorithm performance. While the number of irrelevant data attributes can reduce the level of accuracy of an algorithm. From the results of this study note that the selection feature of information gain ratio can improve the performance of Naive Bayes classification. 

Keywords: Bayes performance improvement, information gain ratio, public
data