This work considers a time series classification task: fault identification in dynamic systems. Two methods are compared: i) Boosting and ii) K-Nearest Neighbors with Dynamic Time Warping distance.