The era of wearable technology and data-driven analytics of physical activities introduces the growing demand for automated classification systems. These systems which categorize sports disciplines as aerobic or anaerobic aim to benefit professional athletes, sports organizations and the ever-growing fitness community. This research aims to develop a machine learning model for classification of sports disciplines and activities. The main objective is to enable the model to make accurate predictions and classify new inputs as aerobic or anaerobic, based on several key features. Trained classification models for this research include Naïve Bayes, Logistic Regression. Support Vector Machine and Decision Tree. These classification models were tested on an observational data collection containing 116 different sports disciplines. The models were compared to determine which one gave the best results for this task. The best performing model, Naïve Bayes, achieved accuracy over 90% across both classes which suggests successful application of a classification model when it comes to determining whether a sports discipline or exercise is primarily aerobic or anaerobic. The results of this research show how machine learning algorithms can help determining the predominant energy system used for which sports discipline. This is especially useful when tailoring personalized endurance, interval and weight loss focused training plans.