Statistical modeling and signal selection in multivariate time series pattern classification

This paper presents an algorithm for selecting a compact subset of relevant signals for pattern classification problems involving multivariate time series (MTS) data. The algorithm uses a statistical causality modeling method to select relevant signals, and a correlation analysis method to remove redundant signals. The MTS signal selection algorithm along with the statistical modeling methods was evaluated through a case study of real-world driving data. From a set of 20 time series signals, the signal selection algorithm selected a subset of 9 signals that are independent and most relevant to the pattern class. We trained a driver state classification system using Random Forest(RF) with the input of 20 original signals, and another system with the selected 9 signals. The experimental results show that the system with 9 selected signals consistently performed better than the system with the original set of 20 signals.