Bug is one of the main concerns of software development for quality assurance. The more complicated system of the software is, the larger the number of bug reports is. Bug-tracking systems manage bug reports to ensure the quality of the software products. The traditional approach of manually tracking and classifying these reports, which wholly depends on the administrators, becomes inefficient due to the increase in system complexity with a tremendous number of bug reports. Furthermore, each bug report contains several attributes including “structured data” and “unstructured data” which need to be carefully examined in order to investigate the insights of the systems. This study focuses on term frequency-inverse document frequency (TF-IDF) with random forest to perform the classification task. Moreover, with an attempt of examining the “unstructured data,” the study also fine-tunes bidirectional encoder representations from transformers (BERT) for text classification. The general accuracy of the model is around 0.79 for the proposed approach, which indicates the efficiency of using BERT in comparison with the previous approach for text classification in bug report analysis.