Real measurement data frequently contains distortions to some extent. In the language of machine learning, these distortions are commonly referred to as noise, and they can result in poor prediction outcomes and decreased classification accuracy. This study investigated the robustness and performance of three machine learning classifiers in the presence of noise. Specifically, four distinct data sets with varied degrees of artificial noise introduced to them were used to train and compare random forests, support vector machines, and artificial neural networks. In conclusion, the random forest classifier outperformed the artificial neural network classifier and was the most robust classifier at most levels. The classifier uses support vector machines to function well at the remaining noise levels. With a linear kernel, the support vector machine classifier proved to be the most dependable and efficient at the two remaining noise levels.