STACIE still contains the synthetic data generation based on a spectral filter that was originally developed to generate synthetic time series for the ACID dataset and is currently still used within STACIE for testing purposes. This approach has known limitations because it produces periodic sequences, of which only half of the generated data is retained to mimic aperiodic behavior.
In ACID 2, synthetic time series will instead be generated directly in the ACID repository by sampling linear stochastic differential equations. This removes the limitations and assumptions described above.
The intention is therefore to remove the synthetic sequence generation from STACIE and refactor the current test cases to rely on the sequence generation of ACID 2. This refactor can only be performed once ACID 2 has been released.
STACIE still contains the synthetic data generation based on a spectral filter that was originally developed to generate synthetic time series for the ACID dataset and is currently still used within STACIE for testing purposes. This approach has known limitations because it produces periodic sequences, of which only half of the generated data is retained to mimic aperiodic behavior.
In ACID 2, synthetic time series will instead be generated directly in the ACID repository by sampling linear stochastic differential equations. This removes the limitations and assumptions described above.
The intention is therefore to remove the synthetic sequence generation from STACIE and refactor the current test cases to rely on the sequence generation of ACID 2. This refactor can only be performed once ACID 2 has been released.