Outlier detection in multivariate functional data through a contaminated mixture model

In an industrial context, the activity of sensors is recorded at a high frequency. A challenge to automatically detect abnormal measurement behavior. Considering sensor measures as functional data, problem can be formulated detection outliers in multivariate data set. Due heterogeneity this set, proposed contaminated mixture model both clusters into homogeneous groups and detects outliers. The main advantage procedure over its competitors that it does not require specify proportion Model inference performed through Expectation-Conditional Maximization algorithm, BIC used select number clusters. Numerical experiments on simulated demonstrate performance achieved by algorithm. particular, outperforms competitors. Its application real which motivated study allows correctly behaviors.