Microgrids have emerged as a new alternative to the traditional growth of transmission and distribution utilities. These power system networks can operate in either an islanded or grid-connected mode, using a combination of renewable and non-renewable energy sources. Their flexibility, efficiency, and cost-effectiveness make them an attractive solution for businesses and organizations, thus helping to move toward a more sustainable energy future. Despite all this, microgrids are prone to various faults that can challenge the stability of the system. Therefore, timely and accurate fault detection and classification are required to maintain reliability. Machine learning has proven its potential as a feasible approach for tackling these issues. This paper focuses on simulating an IEEE 5-bus system using MATLAB/Simulink both in pre-fault and fault conditions to generate data for analysis. The gathered data has been used to train three Python machine learning algorithms: Backpropagation Neural Network, Decision Tree, and KNN. These trained models were utilized for the classification of the type of fault that occurs and the identification of the particular faulted line. Finally, the paper analyzed how each algorithm performs under the change of conditions and determined the most appropriate method for fault detection and classification.