Exploring Interpretable Features for Large Time Series with SE4TeC

Time Series (TS) data are ubiquitous in enormous application fields, such as medicine, multimedia, and finance. In this paper, we present the demonstration with SE4TeC: A Scalable Engine for efficient and expressive Time Series Classification, which is applicable to certain fields in Big Data context, where the TS fea- tures and their extraction process should be interpretable. SE4TeC improves the state of the art solutions by proposing a scalable and highly efficient method to classify TS based on characteris- tic subsequences (i.e., shapelets). We explain the techniques we adopt, and show how to use SE4TeC for exploring the real-life datasets in medical diagnosis and in industrial troubleshooting.