Automated Ticket Classification for Information Technology Helpdesks using Machine Learning

Manual ticket classification is a time-consuming and error-prone process that often leads to misclassification and increases the time it takes to resolve issues. This research investigates using machine learning to classify IT helpdesk tickets automatically. This research uses the GridSearchCV, RandomizedSearchCV, and KFold validation for parameter tuning on three algorithms, KNN, SVM, and Random Forest classifiers, to improve the accuracy of the ticket classifier system. The performance of the classifiers with hyper-tuned parameters was checked against the individual base classifiers using the accuracy score performance metric. KNN and SVM classifiers with hyper-tuned parameters performed better than the corresponding base classifiers. On the contrary, hyperparameter tuning on the Random Forest classifier decreased its prediction accuracy. Overall, the SVM classifier with tuned parameters was the best, with an accuracy score of 94.6%. The dataset was unbalanced, so under-sampling was performed to balance the classes. The results of this study show that machine learning effectively automates ticket classification for IT helpdesks. This can lead to faster resolution times, increased customer satisfaction, and business growth. The study also provides recommendations for future research in evaluating different dataset-balancing approaches.