![]() |
![]() |
![]() |
Download details |
![]() ![]() |
||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||
15 multivariate time series datasets are used to compare S-MTS with competitor algorithms. Most of the studies working on MTS classification follow a different strategy for experimentation which makes the comparison difficult. Some studies evaluates performance using cross-validation. To have fair comparison, we evaluated the performance using both a train/test split and cross-validation. For each dataset, there is train and test file. For cross-validation, train and test data is combined into a single data file. The details about the datasets are provided on Multivariate Time Series Classification with Learned Discretization. S-MTS algorithm requires the datasets to be in certain format. Following table illustrates the format. All time series are concatenated with additional information regarding the time series, time index of the observation and the class of the time series.
You may want to check the source code for data preparation in the SMTS source code to transform the data in to your own format. |
|
|||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||