Project Management Ticket Category Classification Using Machine Learning Method: A Comparative Study

JIRA Software is used to monitor projects and track errors or bugs. Generally, JIRA tickets issued by the operational team consists of 2: EI tickets to report any issues and EACR tickets intended for enhancements or developing new report requesting by clients. However, in its usage, there is often mistakes made in classifying JIRA tickets, which impacts the quality and time delivery to clients. It can be ascertained that the errors were solely human errors, not system errors. To predict categorization errors, we have attempted to apply machine learning using several classifiers and compare the results. From various literature studies, we have tried the 6 most commonly used classifiers for classification and prediction purposes. The classifiers we used are: Multinomial Naive Bayes, Random Forest, Logistic Regression, K-Nearest Neighbors (KNN), Support Vector Machine (SVM), and Extra Trees. Meanwhile, the training and test data comparison was set at a ratio of 70:30. The valuation score from the experiment shows that among other classifier, the Extra Trees classifier produces the highest level of accuracy, F1 score, and recall with values of 0.9950, 0.9950, and 0.9899 respectively. To improve the classifier predictive accuracy, we proposed Extra Trees parameter adjustment.