Review: Mining Recent Temporal Patterns for Event Detection in Multivariate Time Series Data

Iyad Batal et. al. in the paper ”Mining Recent Temporal Patterns for Event Detection in Multivariate Time Series Data” proposed a pattern mining approach for multivariate health data time series which is then used for classification and prediction of diseases. To extract the patterns, they assigned a fuzzy value in time intervals instead of numerical values for each variable. Then, they concatenated several time series of fuzzy values into one sequence which preserved the value of variables rather than aggregating into one value. The authors used this new presentation of the data to mine the recent patterns and fed their classifier. Their results presented the efficiency and accuracy of their method.