benchmark tests for clustering of different types of time series
find good visualization for the bench marking: radar charts? ...
choose some models to run bench marking tests on: mcfly already has 4 types: CNN, LSTM, RestNet, Inception Time
Related to TIME SERIES GENERATION
A. more types of time series
B. be able to generate binary and categorical time series, and mixed typed data
C. being able to generate missing data, even large sections of them, or missing channels
D. being able to generate outlier data values
E. more levels of time series like in between global and local
F. look at existing vocabulary about time series
G. look at auto-encoders and generative models
H. create export facility for generated time series
Like to have:
Related to BENCHMARKING
Related to TIME SERIES GENERATION
A. more types of time series B. be able to generate binary and categorical time series, and mixed typed data C. being able to generate missing data, even large sections of them, or missing channels D. being able to generate outlier data values E. more levels of time series like in between global and local F. look at existing vocabulary about time series G. look at auto-encoders and generative models H. create export facility for generated time series
Probably there is not enough time to do this
Link to article mentioned in Florian's talk