KLASIFIKASI MENGGUNAKAN METODE BAYESIAN BELIEF NETWORKS; CLASSIFICATION TASK USING BAYESIAN BELIEF NETWORKS METHOD

Noviati, Erisa, Sri Haryatmi

2015 | Skripsi | FMIPA

In this modern world nowadays, data mining utilizing become spreading excessively at any various fields and inevitably become part of information technology development. Data mining is the process of extracting meaningful information from a large amount of historical data, through a variety of procedures and methods. There are six main tasks of data mining which is estimation, prediction, classification, clustering, description, and association. Classification become one of the most common data mining tasks because a fairly broad scope of application in such fields as business, economic, health, security, science and technology. One of the classification methods is Bayesian Belief Networks (BBN). BBN method is a method that uses the Bayes principle and theorem assuming that the input variables/attributes can be mutually bound (joint conditionally independent). Then resulted probability value will be used to future data prediction. One of the classification task example in the health field is the process of diagnosis/prognosis/prediction of breast cancer seen from the results of X-rays on the WBCD dataset. Modeling approach of Bayesian Belief Networks with Taboo search algortihm and AIC score function (BBN Taboo-AIC) method is good enough to classify WBCD dataset with the accuracy rate of the training data is 97.9502% and 97.0588% from the test data, with the interpretation of the obtained network results can be easily understood by doctors and health practitioners.

Kata Kunci : data mining; classification; Bayesian Belief Networks.