<![CDATA[PT. Green Air Pacific Surabaya has received an average of 400 - 700 messages per day. So that the incoming email will overlap each other, this can complicate the process of managing email messages that are considered important. This study classifies emails based on their importance using the Naive Bayes method. The word weighting method used is basically TF (Term Frequency) and then normalized to WIDF (Weighted Inverse Document Frequency) weighting. To get a better word index, word weighting is modified by adding word weight if the word is included in the list of predetermined terms. In addition, testing in this study was carried out on the email title, email content and email title with content. From the test results, it was found that the system classifies email data quite well. It can be proven on the highest performance results with an accuracy value of 98,67%, a precision of 100% and an f-measure of 98,99% on the parameters of using email headers, TF weighting, modified TF weighting, and modified WIDF weighting with 240 training data. In addition, the results of testing and data validation using k-fold cross validation also provide an average performance result that is not much different, namely an accuracy value of 95,56%, precision of 93,86% and f-measure of 96,81%..]]>
Copyrights © 2021
